This document: https://colab.research.google.com/drive/12mXX5L1I4rpwl1Jk8hCm-xyAkqiKJEo7?usp=sharing (public link)
Copy of original at https://colab.research.google.com/drive/10BXfzIlkPHrd_zFqIxTc8fm8soa8rYs1 (request access if needed), only editing text cells / manual computation cells to add stats for runs that finished after the original deadline. At the original deadline I only had n=8 (separate transformer training runs, not size of evaluation set) for 3-layer Wu et al 2023 baseline variant on ReCOGS_pos and n=5 for the 4-layer variant; instead of the full n=10,n=10 (which actually completed in the notebook but was not reported in original draft). Here I just update the text cells to add the n=2 and n=5 additional runs for those two to their means/std/confidence intervals reported.
Note this is testing the baseline model, not the RASP model on ReCOGS_pos (same dataset as RASP model), exactly as in Wu et al 2023 (their python scripts) as well as with 3 and 4 layer variants (instead of their original 2).
Note on random seeds used: Wu et al 2023 use the following 5 seeds in their default script 42,66,77,88,99 . When I do n > 5 for a condition, for each subsequent group of 5 runs execute by their script, I just increment their seeds to get the next 5 seeds, e.g. 43,67,78,89,100 , and so on. There was no cherry-picking of seeds.
Wu et al 2023 baseline with ReCOGS LF output format, and fixed positional indices¶
ReCOGS commit used 1b6eca8ff4dca5fd2fb284a7d470998af5083beb
In [ ]:
%cd /content/
!rm -rf ReCOGS
!git clone https://github.com/frankaging/ReCOGS.git
/content Cloning into 'ReCOGS'... remote: Enumerating objects: 436, done. remote: Counting objects: 100% (124/124), done. remote: Compressing objects: 100% (51/51), done. remote: Total 436 (delta 96), reused 92 (delta 73), pack-reused 312 (from 1) Receiving objects: 100% (436/436), 84.71 MiB | 35.84 MiB/s, done. Resolving deltas: 100% (303/303), done. Updating files: 100% (137/137), done.
In [ ]:
%cd ReCOGS
/content/ReCOGS
In [ ]:
!cat README.md
<!-- PROJECT LOGO -->
<br />
<div align="center">
<h3 align="center">ReCOGS: How Incidental Details of a Logical Form Overshadow an Evaluation of Semantic Interpretation</h3>
<p align="center">
Zhengxuan Wu, Christopher D. Manning, Christopher Potts
<br />
<a href="https://arxiv.org/abs/2303.13716"><strong>Read our preprint »</strong></a>
<br />
<br />
<a href="https://github.com/frankaging/ReCOGS/issues">Report Bug</a>
·
<a href="https://nlp.stanford.edu/~wuzhengx/">Contact Us</a>
</p>
</div>
## Introduction
Compositional generalization benchmarks seek to assess whether models can accurately compute **meanings** for novel sentences, but operationalize this in terms of **logical form** (LF) prediction. This raises the concern that semantically irrelevant details of the chosen LFs could shape model performance. We argue that this concern is realized for [the COGS benchmark](https://aclanthology.org/2020.emnlp-main.731.pdf).
## Citation
If you use this repository, please consider to cite our relevant papers:
```stex
@article{wu-etal-2023-recogs,
title={{ReCOGS}: How Incidental Details of a Logical Form Overshadow an Evaluation of Semantic Interpretation},
author={Wu, Zhengxuan and Manning, Christopher D. and Potts, Christopher},
year={2023},
eprint={2303.13716},
archivePrefix={arXiv},
primaryClass={cs.LG}
}
```
## Variants of Reformatted COGS in the Paper
We produce a set of COGS dataset artifacts, each of which is a reformat / transformation of the original COGS dataset. The purpose of these variants is to study how semantic representation of COGS LFs affects model performance. If want to reproduce any of these artifacts, you can simply follow our notebook `second_looks.ipynb`.
#### COGS Token Removal
In this split, we remove redundant tokens in the logical form (e.g., `x`, `_`).
#### COGS Concat
In this split, we concatenate existing examples in the training data to study length generalization v.s. structural generalization.
#### COGS Preposing
In this split, we prepose modifier phrases to study the effects of positional indices in LFs on compositional generalization.
#### COGS Preposing + Sprinkles
In addition to the modification in the previous split, we add in interjections to allow tokens appear in different positions without affecting the semantics.
#### COGS Participle Verb
In this split, we add in an additional semantic parsing rule by augmenting the current training set. Specifically, we add in sentences with participle verbs.
#### COGS Participle Verb (easy)
In addition to the modification in the previous split, we lower the difficulty. Please refer to the paper for details.
#### ReCOGS
For ReCOGS, we try to reduce undesired properties in COGS found in the paper, and enable COGS to measure compositional generalization more *truthfully*.
#### Variable-free COGS
This is a variant we prove in the paper to be *incorrectly* represent the original semantics. We do not recommand to use this variant. This form is proposed by [Qiu et. al., 2022](https://arxiv.org/abs/2112.07610). We use the code released by the original paper to get this form.
## Model Training
### Requirements
- Python 3.6 or 3.7 are supported.
- Pytorch Version: 1.11.0
- Transfermers Version: 4.21.1
- Datasets Version: Version: 2.3.2
### Training **Transformers**
We only have a single training script `run_cogs.py`. You can use it to reproduce our Transformers result. Here is one example,
```bash
python run_cogs.py \
--model_name ende_transformer \
--gpu 1 \
--train_batch_size 128 \
--eval_batch_size 128 \
--lr 0.0001 \
--data_path ./cogs \
--output_dir ./results_cogs \
--lfs cogs \
--do_train \
--do_test \
--do_gen \
--max_seq_len 512 \
--output_json \
--epochs 300 \
--seeds "42;66;77;88;99"
```
### Training **LSTMs**
We only have a single training script `run_cogs.py`. You can use it to reproduce our LSTMs result. Here is one example,
```bash
python run_cogs.py \
--model_name ende_lstm \
--gpu 1 \
--train_batch_size 512 \
--eval_batch_size 256 \
--lr 0.0008 \
--data_path ./cogs \
--output_dir ./results_cogs \
--lfs cogs \
--do_train \
--do_test \
--do_gen \
--max_seq_len 512 \
--output_json \
--epochs 300 \
--seeds "42;66;77;88;99"
```
In [ ]:
!cat /content/ReCOGS/model/encoder_config.json
{
"architectures": [
"Bert"
],
"model_type": "bert",
"attention_probs_dropout_prob": 0.1,
"hidden_act": "gelu",
"hidden_dropout_prob": 0.1,
"hidden_size": 300,
"initializer_range": 0.02,
"intermediate_size": 512,
"num_attention_heads": 4,
"num_hidden_layers": 2,
"type_vocab_size": 2,
"vocab_size": 762,
"pad_token_id": 0,
"bos_token_id": 1,
"eos_token_id": 2,
"unk_token_id": 3,
"mask_token_id": 4,
"cls_token_id": 5,
"sum_token_id": 6,
"nsp_token_id": 7,
"position_embedding_type": "absolute",
"position_embedding_init": "random"
}
In [ ]:
!cat /content/ReCOGS/model/decoder_config.json
{
"architectures": [
"Bert"
],
"model_type": "bert",
"attention_probs_dropout_prob": 0.1,
"hidden_act": "gelu",
"hidden_dropout_prob": 0.1,
"hidden_size": 300,
"initializer_range": 0.02,
"intermediate_size": 512,
"num_attention_heads": 4,
"num_hidden_layers": 2,
"type_vocab_size": 2,
"vocab_size": 729,
"pad_token_id": 0,
"bos_token_id": 1,
"eos_token_id": 2,
"unk_token_id": 3,
"mask_token_id": 4,
"decoder_start_token_id": 1,
"position_embedding_type": "absolute",
"position_embedding_init": "random"
}
In [ ]:
!python run_cogs.py --model_name ende_transformer --gpu 1 --train_batch_size 128 --eval_batch_size 128 --lr 0.0001 --data_path ./recogs_positional_index --output_dir ./results_recogs_positional_index --lfs cogs --do_train --do_test --do_gen --max_seq_len 512 --output_json --epochs 300 --seeds "42;66;77;88;99" # paper's seeds, not sure how they chose them
INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 4344077 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.42 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]We strongly recommend passing in an `attention_mask` since your input_ids may be padded. See https://huggingface.co/docs/transformers/troubleshooting#incorrect-output-when-padding-tokens-arent-masked. /content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:11<00:00, 18.71it/s, loss=5.89] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:10<00:00, 20.35it/s, loss=4.54] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:10<00:00, 20.35it/s, loss=3.54] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:10<00:00, 20.36it/s, loss=2.49] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:10<00:00, 20.37it/s, loss=1.91] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:10<00:00, 20.30it/s, loss=1.58] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:10<00:00, 20.37it/s, loss=1.3] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:10<00:00, 20.37it/s, loss=1.09] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:10<00:00, 20.38it/s, loss=0.97] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0.87] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0.77] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0.68] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0.61] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0.54] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0.47] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0.4] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0.34] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0.29] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0.25] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0.22] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0.19] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0.17] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0.16] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0.14] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0.12] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0.12] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0.11] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0.1] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0.1] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0.08] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0.08] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0.07] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0.07] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0.06] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0.05] Epoch: 35: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 35: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0.05] Epoch: 36: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 36: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0.05] Epoch: 37: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 37: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0.04] Epoch: 38: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 38: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0.04] Epoch: 39: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 39: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0.04] Epoch: 40: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 40: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0.04] Epoch: 41: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 41: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0.03] Epoch: 42: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 42: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0.03] Epoch: 43: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 43: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0.03] Epoch: 44: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 44: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0.04] Epoch: 45: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 45: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0.02] Epoch: 46: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 46: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0.02] Epoch: 47: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 47: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0.02] Epoch: 48: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 48: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0.02] Epoch: 49: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 49: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0.02] Epoch: 50: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 50: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0.02] Epoch: 51: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 51: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0.02] Epoch: 52: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 52: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0.01] Epoch: 53: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 53: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0.01] Epoch: 54: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 54: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0.02] Epoch: 55: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 55: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0.02] Epoch: 56: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 56: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0.02] Epoch: 57: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 57: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0.01] Epoch: 58: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 58: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0.01] Epoch: 59: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 59: 100% 213/213 [00:10<00:00, 20.39it/s, loss=0.01] Epoch: 60: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 60: 100% 213/213 [00:10<00:00, 20.38it/s, loss=0.01] Epoch: 61: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 61: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0.01] Epoch: 62: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 62: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0.01] Epoch: 63: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 63: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0.01] Epoch: 64: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 64: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0.01] Epoch: 65: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 65: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0.01] Epoch: 66: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 66: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0.01] Epoch: 67: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 67: 100% 213/213 [00:10<00:00, 20.38it/s, loss=0.01] Epoch: 68: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 68: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0.01] Epoch: 69: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 69: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0.01] Epoch: 70: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 70: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0.01] Epoch: 71: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 71: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0.01] Epoch: 72: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 72: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0.01] Epoch: 73: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 73: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0.01] Epoch: 74: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 74: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0.01] Epoch: 75: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 75: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0.01] Epoch: 76: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 76: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0.01] Epoch: 77: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 77: 100% 213/213 [00:10<00:00, 20.38it/s, loss=0.01] Epoch: 78: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 78: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0.01] Epoch: 79: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 79: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 80: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 80: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0.01] Epoch: 81: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 81: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 82: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 82: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 83: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 83: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 84: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 84: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 85: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 85: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 86: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 86: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 87: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 87: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0.01] Epoch: 88: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 88: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0.01] Epoch: 89: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 89: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0.01] Epoch: 90: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 90: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 91: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 91: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0.01] Epoch: 92: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 92: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 93: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 93: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 94: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 94: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 95: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 95: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0] Epoch: 96: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 96: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 97: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 97: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 98: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 98: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 99: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 99: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 100: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 100: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 101: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 101: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 102: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 102: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 103: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 103: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 104: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 104: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 105: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 105: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 106: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 106: 100% 213/213 [00:10<00:00, 20.38it/s, loss=0] Epoch: 107: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 107: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 108: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 108: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 109: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 109: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 110: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 110: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 111: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 111: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 112: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 112: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 113: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 113: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 114: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 114: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 115: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 115: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 116: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 116: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 117: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 117: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 118: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 118: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0.01] Epoch: 119: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 119: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 120: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 120: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 121: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 121: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 122: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 122: 100% 213/213 [00:10<00:00, 20.38it/s, loss=0] Epoch: 123: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 123: 100% 213/213 [00:10<00:00, 20.38it/s, loss=0] Epoch: 124: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 124: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 125: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 125: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 126: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 126: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 127: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 127: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 128: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 128: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 129: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 129: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 130: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 130: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 131: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 131: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 132: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 132: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 133: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 133: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 134: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 134: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 135: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 135: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 136: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 136: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 137: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 137: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 138: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 138: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0] Epoch: 139: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 139: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 140: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 140: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 141: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 141: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 142: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 142: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 143: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 143: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0] Epoch: 144: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 144: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 145: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 145: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 146: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 146: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 147: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 147: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 148: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 148: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 149: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 149: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 150: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 150: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 151: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 151: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 152: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 152: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 153: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 153: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 154: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 154: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 155: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 155: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 156: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 156: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 157: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 157: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 158: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 158: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 159: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 159: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 160: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 160: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 161: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 161: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 162: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 162: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 163: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 163: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 164: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 164: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 165: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 165: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 166: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 166: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 167: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 167: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 168: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 168: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 169: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 169: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 170: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 170: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 171: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 171: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 172: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 172: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 173: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 173: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 174: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 174: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 175: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 175: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 176: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 176: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 177: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 177: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 178: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 178: 100% 213/213 [00:10<00:00, 20.39it/s, loss=0] Epoch: 179: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 179: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 180: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 180: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 181: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 181: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 182: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 182: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 183: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 183: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 184: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 184: 100% 213/213 [00:10<00:00, 20.38it/s, loss=0] Epoch: 185: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 185: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 186: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 186: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 187: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 187: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 188: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 188: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 189: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 189: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 190: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 190: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 191: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 191: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 192: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 192: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 193: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 193: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 194: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 194: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 195: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 195: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 196: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 196: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 197: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 197: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 198: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 198: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 199: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 199: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 200: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 200: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 201: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 201: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 202: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 202: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 203: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 203: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 204: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 204: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0] Epoch: 205: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 205: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 206: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 206: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0] Epoch: 207: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 207: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 208: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 208: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 209: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 209: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 210: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 210: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 211: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 211: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 212: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 212: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 213: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 213: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 214: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 214: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 215: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 215: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 216: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 216: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 217: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 217: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 218: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 218: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 219: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 219: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 220: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 220: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 221: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 221: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 222: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 222: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 223: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 223: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 224: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 224: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 225: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 225: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0] Epoch: 226: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 226: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 227: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 227: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 228: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 228: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 229: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 229: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 230: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 230: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 231: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 231: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 232: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 232: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 233: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 233: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 234: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 234: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 235: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 235: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 236: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 236: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0] Epoch: 237: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 237: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 238: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 238: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 239: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 239: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 240: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 240: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 241: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 241: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 242: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 242: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 243: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 243: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 244: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 244: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 245: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 245: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 246: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 246: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 247: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 247: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 248: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 248: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 249: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 249: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 250: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 250: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 251: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 251: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 252: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 252: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 253: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 253: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 254: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 254: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 255: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 255: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 256: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 256: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0] Epoch: 257: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 257: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 258: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 258: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 259: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 259: 100% 213/213 [00:10<00:00, 20.38it/s, loss=0] Epoch: 260: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 260: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 261: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 261: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 262: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 262: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 263: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 263: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 264: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 264: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 265: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 265: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 266: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 266: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 267: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 267: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 268: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 268: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 269: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 269: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 270: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 270: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0] Epoch: 271: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 271: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 272: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 272: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 273: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 273: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 274: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 274: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 275: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 275: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 276: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 276: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 277: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 277: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 278: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 278: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 279: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 279: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 280: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 280: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 281: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 281: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 282: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 282: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 283: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 283: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 284: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 284: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 285: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 285: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 286: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 286: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 287: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 287: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 288: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 288: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 289: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 289: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 290: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 290: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 291: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 291: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 292: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 292: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 293: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 293: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0] Epoch: 294: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 294: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 295: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 295: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 296: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 296: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 297: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 297: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 298: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 298: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 299: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 299: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 100% 300/300 [52:43<00:00, 10.55s/it] INFO:root:Training is finished ... Iteration: 100% 24/24 [00:10<00:00, 2.35it/s, acc=1] Iteration: 100% 165/165 [20:40<00:00, 7.52s/it, acc=0.874] obj_pp_to_subj_pp: 14.8 cp_recursion: 52.2 pp_recursion: 43.9 subj_to_obj_proper: 95.5 prim_to_obj_proper: 95.5 prim_to_subj_proper: 99.9 LEX: 95.52000000000001 OVERALL: 87.36190476190477 INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 4344077 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.66 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:10<00:00, 20.18it/s, loss=5.99] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:10<00:00, 20.32it/s, loss=4.61] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:10<00:00, 20.29it/s, loss=3.61] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:10<00:00, 20.29it/s, loss=2.53] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:10<00:00, 20.31it/s, loss=1.96] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:10<00:00, 20.23it/s, loss=1.6] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:10<00:00, 20.31it/s, loss=1.32] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:10<00:00, 20.29it/s, loss=1.13] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:10<00:00, 20.27it/s, loss=1] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0.9] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0.8] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0.71] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0.63] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0.54] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0.47] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0.41] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0.36] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0.31] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0.27] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0.23] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0.21] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0.17] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0.15] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0.13] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0.12] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0.11] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0.11] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0.09] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0.09] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0.08] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0.07] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0.06] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0.05] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0.05] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0.05] Epoch: 35: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 35: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0.04] Epoch: 36: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 36: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0.04] Epoch: 37: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 37: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0.04] Epoch: 38: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 38: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0.04] Epoch: 39: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 39: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0.03] Epoch: 40: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 40: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0.03] Epoch: 41: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 41: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0.03] Epoch: 42: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 42: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0.03] Epoch: 43: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 43: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0.03] Epoch: 44: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 44: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0.02] Epoch: 45: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 45: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0.02] Epoch: 46: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 46: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0.02] Epoch: 47: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 47: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0.02] Epoch: 48: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 48: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0.02] Epoch: 49: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 49: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0.02] Epoch: 50: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 50: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0.02] Epoch: 51: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 51: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0.03] Epoch: 52: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 52: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0.02] Epoch: 53: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 53: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0.01] Epoch: 54: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 54: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0.02] Epoch: 55: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 55: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0.01] Epoch: 56: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 56: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0.01] Epoch: 57: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 57: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0.01] Epoch: 58: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 58: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0.01] Epoch: 59: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 59: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0.01] Epoch: 60: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 60: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0.01] Epoch: 61: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 61: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0.01] Epoch: 62: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 62: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0.01] Epoch: 63: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 63: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0.01] Epoch: 64: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 64: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0.01] Epoch: 65: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 65: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0.01] Epoch: 66: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 66: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0.01] Epoch: 67: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 67: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0.01] Epoch: 68: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 68: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0.01] Epoch: 69: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 69: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 70: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 70: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0.01] Epoch: 71: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 71: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0.01] Epoch: 72: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 72: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0.01] Epoch: 73: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 73: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0.01] Epoch: 74: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 74: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0.01] Epoch: 75: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 75: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0.01] Epoch: 76: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 76: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0.01] Epoch: 77: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 77: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0.01] Epoch: 78: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 78: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 79: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 79: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0.01] Epoch: 80: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 80: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 81: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 81: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0.01] Epoch: 82: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 82: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0.01] Epoch: 83: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 83: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0.01] Epoch: 84: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 84: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 85: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 85: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 86: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 86: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0.01] Epoch: 87: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 87: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 88: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 88: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0.01] Epoch: 89: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 89: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0] Epoch: 90: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 90: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 91: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 91: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 92: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 92: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 93: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 93: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 94: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 94: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 95: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 95: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0.01] Epoch: 96: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 96: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 97: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 97: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 98: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 98: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 99: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 99: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0.01] Epoch: 100: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 100: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 101: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 101: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 102: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 102: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 103: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 103: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 104: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 104: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 105: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 105: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 106: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 106: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 107: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 107: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 108: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 108: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 109: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 109: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 110: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 110: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 111: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 111: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 112: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 112: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 113: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 113: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0] Epoch: 114: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 114: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 115: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 115: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 116: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 116: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 117: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 117: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 118: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 118: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0] Epoch: 119: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 119: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 120: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 120: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 121: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 121: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 122: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 122: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 123: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 123: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 124: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 124: 100% 213/213 [00:10<00:00, 20.38it/s, loss=0] Epoch: 125: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 125: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 126: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 126: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 127: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 127: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 128: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 128: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 129: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 129: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 130: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 130: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 131: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 131: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 132: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 132: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 133: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 133: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 134: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 134: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 135: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 135: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 136: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 136: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 137: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 137: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 138: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 138: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0] Epoch: 139: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 139: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 140: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 140: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 141: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 141: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 142: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 142: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 143: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 143: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 144: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 144: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 145: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 145: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 146: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 146: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 147: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 147: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 148: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 148: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 149: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 149: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 150: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 150: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 151: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 151: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 152: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 152: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 153: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 153: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 154: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 154: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 155: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 155: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 156: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 156: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 157: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 157: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 158: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 158: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 159: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 159: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 160: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 160: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 161: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 161: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 162: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 162: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 163: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 163: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 164: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 164: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 165: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 165: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 166: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 166: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 167: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 167: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 168: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 168: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 169: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 169: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 170: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 170: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0] Epoch: 171: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 171: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 172: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 172: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 173: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 173: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 174: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 174: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 175: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 175: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 176: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 176: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 177: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 177: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 178: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 178: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 179: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 179: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 180: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 180: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 181: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 181: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 182: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 182: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 183: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 183: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 184: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 184: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 185: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 185: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 186: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 186: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 187: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 187: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 188: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 188: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 189: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 189: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 190: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 190: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 191: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 191: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 192: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 192: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 193: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 193: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 194: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 194: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 195: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 195: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 196: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 196: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 197: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 197: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 198: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 198: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 199: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 199: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0] Epoch: 200: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 200: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 201: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 201: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 202: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 202: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 203: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 203: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 204: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 204: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 205: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 205: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 206: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 206: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 207: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 207: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 208: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 208: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 209: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 209: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 210: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 210: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 211: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 211: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 212: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 212: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 213: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 213: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 214: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 214: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 215: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 215: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 216: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 216: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 217: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 217: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 218: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 218: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 219: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 219: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 220: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 220: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 221: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 221: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 222: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 222: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 223: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 223: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 224: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 224: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 225: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 225: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 226: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 226: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 227: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 227: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 228: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 228: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 229: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 229: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 230: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 230: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 231: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 231: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 232: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 232: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 233: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 233: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0] Epoch: 234: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 234: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 235: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 235: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 236: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 236: 100% 213/213 [00:10<00:00, 20.40it/s, loss=0] Epoch: 237: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 237: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0] Epoch: 238: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 238: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 239: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 239: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 240: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 240: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 241: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 241: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 242: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 242: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 243: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 243: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 244: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 244: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 245: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 245: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 246: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 246: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 247: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 247: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 248: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 248: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 249: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 249: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 250: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 250: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 251: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 251: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 252: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 252: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 253: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 253: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 254: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 254: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 255: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 255: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 256: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 256: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 257: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 257: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 258: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 258: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 259: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 259: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 260: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 260: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 261: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 261: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 262: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 262: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 263: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 263: 100% 213/213 [00:10<00:00, 20.38it/s, loss=0] Epoch: 264: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 264: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 265: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 265: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 266: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 266: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 267: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 267: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 268: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 268: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 269: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 269: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 270: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 270: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 271: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 271: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 272: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 272: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 273: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 273: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 274: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 274: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 275: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 275: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 276: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 276: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 277: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 277: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 278: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 278: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 279: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 279: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 280: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 280: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 281: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 281: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 282: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 282: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 283: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 283: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 284: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 284: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 285: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 285: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 286: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 286: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 287: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 287: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 288: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 288: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 289: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 289: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 290: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 290: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 291: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 291: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 292: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 292: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 293: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 293: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 294: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 294: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 295: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 295: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 296: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 296: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 297: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 297: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 298: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 298: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 299: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 299: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 100% 300/300 [52:42<00:00, 10.54s/it] INFO:root:Training is finished ... Iteration: 100% 24/24 [00:10<00:00, 2.27it/s, acc=1] Iteration: 100% 165/165 [20:16<00:00, 7.38s/it, acc=0.898] obj_pp_to_subj_pp: 19.7 cp_recursion: 53.6 pp_recursion: 42.4 subj_to_obj_proper: 90.5 prim_to_obj_proper: 86.1 prim_to_subj_proper: 100.0 LEX: 99.63333333333333 OVERALL: 89.84761904761905 INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 4344077 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.77 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:10<00:00, 20.18it/s, loss=6.02] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:10<00:00, 20.35it/s, loss=4.68] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:10<00:00, 20.35it/s, loss=3.62] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:10<00:00, 20.29it/s, loss=2.5] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:10<00:00, 20.40it/s, loss=1.96] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:10<00:00, 20.35it/s, loss=1.62] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:10<00:00, 20.36it/s, loss=1.36] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:10<00:00, 20.32it/s, loss=1.18] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:10<00:00, 20.37it/s, loss=1.06] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0.95] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0.86] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0.77] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0.69] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0.62] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0.55] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:10<00:00, 20.38it/s, loss=0.49] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0.43] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0.37] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0.32] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0.29] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0.25] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0.22] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0.2] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0.18] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:10<00:00, 20.38it/s, loss=0.16] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0.14] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0.13] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0.11] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0.11] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0.1] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0.1] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0.08] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0.07] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0.07] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0.07] Epoch: 35: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 35: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0.06] Epoch: 36: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 36: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0.06] Epoch: 37: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 37: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0.05] Epoch: 38: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 38: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0.05] Epoch: 39: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 39: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0.04] Epoch: 40: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 40: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0.05] Epoch: 41: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 41: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0.04] Epoch: 42: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 42: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0.03] Epoch: 43: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 43: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0.04] Epoch: 44: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 44: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0.03] Epoch: 45: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 45: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0.03] Epoch: 46: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 46: 100% 213/213 [00:10<00:00, 20.38it/s, loss=0.03] Epoch: 47: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 47: 100% 213/213 [00:10<00:00, 20.38it/s, loss=0.03] Epoch: 48: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 48: 100% 213/213 [00:10<00:00, 20.38it/s, loss=0.02] Epoch: 49: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 49: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0.02] Epoch: 50: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 50: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0.03] Epoch: 51: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 51: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0.02] Epoch: 52: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 52: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0.03] Epoch: 53: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 53: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0.02] Epoch: 54: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 54: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0.02] Epoch: 55: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 55: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0.02] Epoch: 56: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 56: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0.02] Epoch: 57: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 57: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0.02] Epoch: 58: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 58: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0.02] Epoch: 59: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 59: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0.02] Epoch: 60: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 60: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0.02] Epoch: 61: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 61: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0.01] Epoch: 62: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 62: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0.01] Epoch: 63: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 63: 100% 213/213 [00:10<00:00, 20.39it/s, loss=0.02] Epoch: 64: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 64: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0.02] Epoch: 65: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 65: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0.01] Epoch: 66: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 66: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0.01] Epoch: 67: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 67: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0.01] Epoch: 68: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 68: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0.01] Epoch: 69: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 69: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0.01] Epoch: 70: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 70: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0.01] Epoch: 71: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 71: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0.01] Epoch: 72: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 72: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0.01] Epoch: 73: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 73: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0.01] Epoch: 74: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 74: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0.02] Epoch: 75: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 75: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0.01] Epoch: 76: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 76: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0.01] Epoch: 77: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 77: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0.01] Epoch: 78: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 78: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0.01] Epoch: 79: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 79: 100% 213/213 [00:10<00:00, 20.38it/s, loss=0.01] Epoch: 80: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 80: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0.01] Epoch: 81: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 81: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0.01] Epoch: 82: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 82: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0.01] Epoch: 83: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 83: 100% 213/213 [00:10<00:00, 20.38it/s, loss=0.01] Epoch: 84: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 84: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0.01] Epoch: 85: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 85: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0.01] Epoch: 86: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 86: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0.01] Epoch: 87: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 87: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 88: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 88: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 89: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 89: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0.01] Epoch: 90: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 90: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0.01] Epoch: 91: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 91: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 92: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 92: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0.01] Epoch: 93: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 93: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 94: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 94: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0.01] Epoch: 95: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 95: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 96: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 96: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 97: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 97: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 98: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 98: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 99: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 99: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 100: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 100: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 101: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 101: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0.01] Epoch: 102: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 102: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 103: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 103: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 104: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 104: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 105: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 105: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 106: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 106: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 107: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 107: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 108: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 108: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 109: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 109: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 110: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 110: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 111: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 111: 100% 213/213 [00:10<00:00, 20.39it/s, loss=0] Epoch: 112: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 112: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 113: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 113: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 114: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 114: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 115: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 115: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 116: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 116: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 117: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 117: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 118: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 118: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0] Epoch: 119: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 119: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0.01] Epoch: 120: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 120: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 121: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 121: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 122: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 122: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 123: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 123: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 124: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 124: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 125: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 125: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 126: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 126: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 127: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 127: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 128: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 128: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0.01] Epoch: 129: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 129: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 130: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 130: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 131: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 131: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 132: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 132: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 133: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 133: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 134: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 134: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 135: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 135: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 136: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 136: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 137: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 137: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 138: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 138: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 139: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 139: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0.01] Epoch: 140: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 140: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 141: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 141: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 142: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 142: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 143: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 143: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 144: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 144: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 145: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 145: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 146: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 146: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 147: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 147: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 148: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 148: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 149: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 149: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 150: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 150: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 151: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 151: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 152: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 152: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 153: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 153: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 154: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 154: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 155: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 155: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 156: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 156: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 157: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 157: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 158: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 158: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 159: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 159: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0.01] Epoch: 160: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 160: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 161: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 161: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 162: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 162: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 163: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 163: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 164: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 164: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 165: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 165: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 166: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 166: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 167: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 167: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 168: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 168: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 169: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 169: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 170: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 170: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 171: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 171: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0] Epoch: 172: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 172: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 173: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 173: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 174: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 174: 100% 213/213 [00:10<00:00, 20.38it/s, loss=0] Epoch: 175: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 175: 100% 213/213 [00:10<00:00, 20.38it/s, loss=0] Epoch: 176: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 176: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 177: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 177: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 178: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 178: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 179: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 179: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 180: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 180: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 181: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 181: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 182: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 182: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 183: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 183: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 184: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 184: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 185: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 185: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 186: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 186: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 187: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 187: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 188: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 188: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 189: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 189: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 190: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 190: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 191: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 191: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 192: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 192: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 193: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 193: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 194: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 194: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 195: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 195: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 196: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 196: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 197: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 197: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 198: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 198: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 199: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 199: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 200: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 200: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 201: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 201: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 202: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 202: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 203: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 203: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 204: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 204: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 205: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 205: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 206: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 206: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 207: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 207: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 208: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 208: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 209: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 209: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 210: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 210: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 211: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 211: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0] Epoch: 212: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 212: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 213: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 213: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 214: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 214: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 215: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 215: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 216: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 216: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 217: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 217: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 218: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 218: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 219: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 219: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 220: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 220: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 221: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 221: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 222: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 222: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0] Epoch: 223: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 223: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 224: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 224: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 225: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 225: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 226: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 226: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0] Epoch: 227: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 227: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 228: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 228: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 229: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 229: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 230: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 230: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 231: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 231: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0] Epoch: 232: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 232: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 233: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 233: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 234: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 234: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 235: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 235: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 236: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 236: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 237: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 237: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 238: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 238: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 239: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 239: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 240: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 240: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 241: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 241: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 242: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 242: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 243: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 243: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 244: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 244: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0] Epoch: 245: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 245: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0] Epoch: 246: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 246: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 247: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 247: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 248: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 248: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 249: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 249: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 250: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 250: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 251: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 251: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 252: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 252: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 253: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 253: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 254: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 254: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 255: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 255: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 256: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 256: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 257: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 257: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0] Epoch: 258: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 258: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 259: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 259: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 260: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 260: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 261: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 261: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 262: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 262: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 263: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 263: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 264: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 264: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 265: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 265: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 266: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 266: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 267: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 267: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 268: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 268: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 269: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 269: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 270: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 270: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 271: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 271: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 272: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 272: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 273: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 273: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 274: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 274: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 275: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 275: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 276: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 276: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 277: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 277: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 278: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 278: 100% 213/213 [00:10<00:00, 20.38it/s, loss=0] Epoch: 279: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 279: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 280: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 280: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 281: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 281: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 282: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 282: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 283: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 283: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 284: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 284: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 285: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 285: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 286: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 286: 100% 213/213 [00:10<00:00, 20.38it/s, loss=0] Epoch: 287: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 287: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 288: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 288: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 289: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 289: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 290: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 290: 100% 213/213 [00:10<00:00, 20.38it/s, loss=0] Epoch: 291: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 291: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 292: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 292: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 293: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 293: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 294: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 294: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0] Epoch: 295: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 295: 100% 213/213 [00:10<00:00, 20.38it/s, loss=0] Epoch: 296: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 296: 100% 213/213 [00:10<00:00, 20.35it/s, loss=0] Epoch: 297: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 297: 100% 213/213 [00:10<00:00, 20.34it/s, loss=0] Epoch: 298: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 298: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 299: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 299: 100% 213/213 [00:10<00:00, 20.36it/s, loss=0] Epoch: 100% 300/300 [52:42<00:00, 10.54s/it] INFO:root:Training is finished ... Iteration: 100% 24/24 [00:10<00:00, 2.28it/s, acc=1] Iteration: 100% 165/165 [17:31<00:00, 6.38s/it, acc=0.901] obj_pp_to_subj_pp: 31.0 cp_recursion: 53.7 pp_recursion: 43.1 subj_to_obj_proper: 88.0 prim_to_obj_proper: 92.4 prim_to_subj_proper: 99.9 LEX: 98.99333333333334 OVERALL: 90.14285714285715 INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 4344077 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.88 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:10<00:00, 20.25it/s, loss=5.98] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:10<00:00, 20.48it/s, loss=4.6] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:10<00:00, 20.46it/s, loss=3.51] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:10<00:00, 20.47it/s, loss=2.38] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:10<00:00, 20.43it/s, loss=1.81] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:10<00:00, 20.47it/s, loss=1.48] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:10<00:00, 20.42it/s, loss=1.2] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0.99] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0.86] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0.77] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:10<00:00, 20.38it/s, loss=0.68] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0.58] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:10<00:00, 20.49it/s, loss=0.51] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0.43] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0.36] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0.3] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:10<00:00, 20.41it/s, loss=0.25] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0.19] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0.16] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0.12] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0.1] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0.08] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:10<00:00, 20.43it/s, loss=0.07] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:10<00:00, 20.49it/s, loss=0.06] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0.04] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0.04] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0.03] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:10<00:00, 20.39it/s, loss=0.03] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0.03] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0.02] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:10<00:00, 20.44it/s, loss=0.02] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0.01] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0.01] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:10<00:00, 20.40it/s, loss=0.01] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0.01] Epoch: 35: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 35: 100% 213/213 [00:10<00:00, 20.49it/s, loss=0.01] Epoch: 36: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 36: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0.01] Epoch: 37: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 37: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0.01] Epoch: 38: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 38: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0.01] Epoch: 39: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 39: 100% 213/213 [00:10<00:00, 20.44it/s, loss=0.01] Epoch: 40: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 40: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0] Epoch: 41: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 41: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 42: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 42: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0.01] Epoch: 43: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 43: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0.01] Epoch: 44: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 44: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 45: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 45: 100% 213/213 [00:10<00:00, 20.41it/s, loss=0] Epoch: 46: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 46: 100% 213/213 [00:10<00:00, 20.49it/s, loss=0.01] Epoch: 47: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 47: 100% 213/213 [00:10<00:00, 20.49it/s, loss=0] Epoch: 48: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 48: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0.01] Epoch: 49: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 49: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 50: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 50: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 51: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 51: 100% 213/213 [00:10<00:00, 20.45it/s, loss=0] Epoch: 52: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 52: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 53: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 53: 100% 213/213 [00:10<00:00, 20.42it/s, loss=0] Epoch: 54: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 54: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 55: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 55: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 56: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 56: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0] Epoch: 57: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 57: 100% 213/213 [00:10<00:00, 20.41it/s, loss=0] Epoch: 58: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 58: 100% 213/213 [00:10<00:00, 20.49it/s, loss=0] Epoch: 59: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 59: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 60: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 60: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 61: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 61: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 62: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 62: 100% 213/213 [00:10<00:00, 20.49it/s, loss=0] Epoch: 63: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 63: 100% 213/213 [00:10<00:00, 20.44it/s, loss=0] Epoch: 64: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 64: 100% 213/213 [00:10<00:00, 20.49it/s, loss=0] Epoch: 65: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 65: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 66: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 66: 100% 213/213 [00:10<00:00, 20.49it/s, loss=0] Epoch: 67: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 67: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0] Epoch: 68: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 68: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 69: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 69: 100% 213/213 [00:10<00:00, 20.45it/s, loss=0] Epoch: 70: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 70: 100% 213/213 [00:10<00:00, 20.49it/s, loss=0] Epoch: 71: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 71: 100% 213/213 [00:10<00:00, 20.44it/s, loss=0] Epoch: 72: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 72: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 73: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 73: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 74: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 74: 100% 213/213 [00:10<00:00, 20.44it/s, loss=0] Epoch: 75: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 75: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 76: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 76: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 77: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 77: 100% 213/213 [00:10<00:00, 20.42it/s, loss=0] Epoch: 78: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 78: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 79: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 79: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 80: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 80: 100% 213/213 [00:10<00:00, 20.41it/s, loss=0] Epoch: 81: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 81: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 82: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 82: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 83: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 83: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 84: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 84: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 85: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 85: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 86: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 86: 100% 213/213 [00:10<00:00, 20.44it/s, loss=0] Epoch: 87: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 87: 100% 213/213 [00:10<00:00, 20.45it/s, loss=0] Epoch: 88: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 88: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 89: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 89: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 90: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 90: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0] Epoch: 91: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 91: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 92: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 92: 100% 213/213 [00:10<00:00, 20.40it/s, loss=0] Epoch: 93: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 93: 100% 213/213 [00:10<00:00, 20.49it/s, loss=0] Epoch: 94: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 94: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 95: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 95: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 96: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 96: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 97: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 97: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 98: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 98: 100% 213/213 [00:10<00:00, 20.38it/s, loss=0] Epoch: 99: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 99: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 100: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 100: 100% 213/213 [00:10<00:00, 20.43it/s, loss=0] Epoch: 101: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 101: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 102: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 102: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 103: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 103: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 104: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 104: 100% 213/213 [00:10<00:00, 20.40it/s, loss=0] Epoch: 105: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 105: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0] Epoch: 106: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 106: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0] Epoch: 107: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 107: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 108: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 108: 100% 213/213 [00:10<00:00, 20.45it/s, loss=0] Epoch: 109: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 109: 100% 213/213 [00:10<00:00, 20.49it/s, loss=0] Epoch: 110: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 110: 100% 213/213 [00:10<00:00, 20.41it/s, loss=0] Epoch: 111: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 111: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0] Epoch: 112: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 112: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 113: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 113: 100% 213/213 [00:10<00:00, 20.49it/s, loss=0] Epoch: 114: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 114: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 115: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 115: 100% 213/213 [00:10<00:00, 20.40it/s, loss=0] Epoch: 116: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 116: 100% 213/213 [00:10<00:00, 20.49it/s, loss=0] Epoch: 117: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 117: 100% 213/213 [00:10<00:00, 20.49it/s, loss=0] Epoch: 118: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 118: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 119: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 119: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 120: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 120: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 121: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 121: 100% 213/213 [00:10<00:00, 20.42it/s, loss=0] Epoch: 122: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 122: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 123: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 123: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0] Epoch: 124: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 124: 100% 213/213 [00:10<00:00, 20.44it/s, loss=0] Epoch: 125: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 125: 100% 213/213 [00:10<00:00, 20.49it/s, loss=0] Epoch: 126: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 126: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 127: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 127: 100% 213/213 [00:10<00:00, 20.41it/s, loss=0] Epoch: 128: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 128: 100% 213/213 [00:10<00:00, 20.49it/s, loss=0] Epoch: 129: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 129: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 130: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 130: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 131: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 131: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 132: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 132: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0] Epoch: 133: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 133: 100% 213/213 [00:10<00:00, 20.40it/s, loss=0] Epoch: 134: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 134: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0] Epoch: 135: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 135: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0] Epoch: 136: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 136: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 137: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 137: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0] Epoch: 138: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 138: 100% 213/213 [00:10<00:00, 20.45it/s, loss=0] Epoch: 139: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 139: 100% 213/213 [00:10<00:00, 20.40it/s, loss=0] Epoch: 140: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 140: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 141: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 141: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 142: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 142: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 143: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 143: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 144: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 144: 100% 213/213 [00:10<00:00, 20.51it/s, loss=0] Epoch: 145: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 145: 100% 213/213 [00:10<00:00, 20.45it/s, loss=0] Epoch: 146: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 146: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 147: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 147: 100% 213/213 [00:10<00:00, 20.45it/s, loss=0] Epoch: 148: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 148: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 149: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 149: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 150: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 150: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 151: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 151: 100% 213/213 [00:10<00:00, 20.42it/s, loss=0] Epoch: 152: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 152: 100% 213/213 [00:10<00:00, 20.49it/s, loss=0] Epoch: 153: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 153: 100% 213/213 [00:10<00:00, 20.49it/s, loss=0] Epoch: 154: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 154: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 155: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 155: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 156: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 156: 100% 213/213 [00:10<00:00, 20.49it/s, loss=0] Epoch: 157: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 157: 100% 213/213 [00:10<00:00, 20.40it/s, loss=0] Epoch: 158: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 158: 100% 213/213 [00:10<00:00, 20.45it/s, loss=0] Epoch: 159: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 159: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 160: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 160: 100% 213/213 [00:10<00:00, 20.49it/s, loss=0] Epoch: 161: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 161: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 162: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 162: 100% 213/213 [00:10<00:00, 20.41it/s, loss=0] Epoch: 163: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 163: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0] Epoch: 164: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 164: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 165: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 165: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 166: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 166: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0] Epoch: 167: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 167: 100% 213/213 [00:10<00:00, 20.49it/s, loss=0] Epoch: 168: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 168: 100% 213/213 [00:10<00:00, 20.39it/s, loss=0] Epoch: 169: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 169: 100% 213/213 [00:10<00:00, 20.43it/s, loss=0] Epoch: 170: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 170: 100% 213/213 [00:10<00:00, 20.44it/s, loss=0] Epoch: 171: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 171: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0] Epoch: 172: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 172: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 173: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 173: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 174: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 174: 100% 213/213 [00:10<00:00, 20.42it/s, loss=0] Epoch: 175: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 175: 100% 213/213 [00:10<00:00, 20.50it/s, loss=0] Epoch: 176: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 176: 100% 213/213 [00:10<00:00, 20.49it/s, loss=0] Epoch: 177: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 177: 100% 213/213 [00:10<00:00, 20.45it/s, loss=0] Epoch: 178: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 178: 100% 213/213 [00:10<00:00, 20.44it/s, loss=0] Epoch: 179: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 179: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0] Epoch: 180: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 180: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 181: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 181: 100% 213/213 [00:10<00:00, 20.44it/s, loss=0] Epoch: 182: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 182: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 183: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 183: 100% 213/213 [00:10<00:00, 20.50it/s, loss=0] Epoch: 184: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 184: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 185: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 185: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 186: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 186: 100% 213/213 [00:10<00:00, 20.42it/s, loss=0] Epoch: 187: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 187: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0] Epoch: 188: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 188: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0] Epoch: 189: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 189: 100% 213/213 [00:10<00:00, 20.50it/s, loss=0] Epoch: 190: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 190: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 191: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 191: 100% 213/213 [00:10<00:00, 20.49it/s, loss=0] Epoch: 192: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 192: 100% 213/213 [00:10<00:00, 20.42it/s, loss=0] Epoch: 193: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 193: 100% 213/213 [00:10<00:00, 20.45it/s, loss=0] Epoch: 194: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 194: 100% 213/213 [00:10<00:00, 20.41it/s, loss=0] Epoch: 195: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 195: 100% 213/213 [00:10<00:00, 20.49it/s, loss=0] Epoch: 196: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 196: 100% 213/213 [00:10<00:00, 20.43it/s, loss=0] Epoch: 197: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 197: 100% 213/213 [00:10<00:00, 20.45it/s, loss=0] Epoch: 198: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 198: 100% 213/213 [00:10<00:00, 20.39it/s, loss=0] Epoch: 199: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 199: 100% 213/213 [00:10<00:00, 20.42it/s, loss=0] Epoch: 200: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 200: 100% 213/213 [00:10<00:00, 20.40it/s, loss=0] Epoch: 201: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 201: 100% 213/213 [00:10<00:00, 20.44it/s, loss=0] Epoch: 202: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 202: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 203: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 203: 100% 213/213 [00:10<00:00, 20.45it/s, loss=0] Epoch: 204: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 204: 100% 213/213 [00:10<00:00, 20.37it/s, loss=0] Epoch: 205: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 205: 100% 213/213 [00:10<00:00, 20.45it/s, loss=0] Epoch: 206: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 206: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0] Epoch: 207: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 207: 100% 213/213 [00:10<00:00, 20.45it/s, loss=0] Epoch: 208: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 208: 100% 213/213 [00:10<00:00, 20.44it/s, loss=0] Epoch: 209: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 209: 100% 213/213 [00:10<00:00, 20.42it/s, loss=0] Epoch: 210: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 210: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 211: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 211: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 212: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 212: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 213: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 213: 100% 213/213 [00:10<00:00, 20.44it/s, loss=0] Epoch: 214: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 214: 100% 213/213 [00:10<00:00, 20.49it/s, loss=0] Epoch: 215: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 215: 100% 213/213 [00:10<00:00, 20.40it/s, loss=0] Epoch: 216: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 216: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 217: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 217: 100% 213/213 [00:10<00:00, 20.43it/s, loss=0] Epoch: 218: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 218: 100% 213/213 [00:10<00:00, 20.50it/s, loss=0] Epoch: 219: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 219: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 220: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 220: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 221: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 221: 100% 213/213 [00:10<00:00, 20.40it/s, loss=0] Epoch: 222: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 222: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 223: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 223: 100% 213/213 [00:10<00:00, 20.44it/s, loss=0] Epoch: 224: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 224: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 225: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 225: 100% 213/213 [00:10<00:00, 20.44it/s, loss=0] Epoch: 226: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 226: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0] Epoch: 227: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 227: 100% 213/213 [00:10<00:00, 20.38it/s, loss=0] Epoch: 228: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 228: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 229: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 229: 100% 213/213 [00:10<00:00, 20.45it/s, loss=0] Epoch: 230: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 230: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 231: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 231: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0] Epoch: 232: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 232: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 233: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 233: 100% 213/213 [00:10<00:00, 20.39it/s, loss=0] Epoch: 234: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 234: 100% 213/213 [00:10<00:00, 20.49it/s, loss=0] Epoch: 235: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 235: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 236: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 236: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0] Epoch: 237: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 237: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 238: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 238: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0] Epoch: 239: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 239: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0] Epoch: 240: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 240: 100% 213/213 [00:10<00:00, 20.45it/s, loss=0] Epoch: 241: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 241: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 242: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 242: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 243: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 243: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 244: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 244: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 245: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 245: 100% 213/213 [00:10<00:00, 20.40it/s, loss=0] Epoch: 246: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 246: 100% 213/213 [00:10<00:00, 20.50it/s, loss=0] Epoch: 247: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 247: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 248: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 248: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 249: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 249: 100% 213/213 [00:10<00:00, 20.51it/s, loss=0] Epoch: 250: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 250: 100% 213/213 [00:10<00:00, 20.40it/s, loss=0] Epoch: 251: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 251: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 252: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 252: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 253: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 253: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 254: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 254: 100% 213/213 [00:10<00:00, 20.49it/s, loss=0] Epoch: 255: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 255: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 256: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 256: 100% 213/213 [00:10<00:00, 20.39it/s, loss=0] Epoch: 257: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 257: 100% 213/213 [00:10<00:00, 20.51it/s, loss=0] Epoch: 258: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 258: 100% 213/213 [00:10<00:00, 20.49it/s, loss=0] Epoch: 259: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 259: 100% 213/213 [00:10<00:00, 20.45it/s, loss=0] Epoch: 260: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 260: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 261: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 261: 100% 213/213 [00:10<00:00, 20.49it/s, loss=0] Epoch: 262: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 262: 100% 213/213 [00:10<00:00, 20.40it/s, loss=0] Epoch: 263: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 263: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 264: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 264: 100% 213/213 [00:10<00:00, 20.45it/s, loss=0] Epoch: 265: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 265: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 266: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 266: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0] Epoch: 267: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 267: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 268: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 268: 100% 213/213 [00:10<00:00, 20.43it/s, loss=0] Epoch: 269: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 269: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 270: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 270: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 271: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 271: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 272: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 272: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0] Epoch: 273: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 273: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 274: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 274: 100% 213/213 [00:10<00:00, 20.42it/s, loss=0] Epoch: 275: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 275: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0] Epoch: 276: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 276: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 277: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 277: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 278: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 278: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0] Epoch: 279: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 279: 100% 213/213 [00:10<00:00, 20.49it/s, loss=0] Epoch: 280: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 280: 100% 213/213 [00:10<00:00, 20.42it/s, loss=0] Epoch: 281: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 281: 100% 213/213 [00:10<00:00, 20.49it/s, loss=0] Epoch: 282: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 282: 100% 213/213 [00:10<00:00, 20.45it/s, loss=0] Epoch: 283: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 283: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0] Epoch: 284: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 284: 100% 213/213 [00:10<00:00, 20.49it/s, loss=0] Epoch: 285: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 285: 100% 213/213 [00:10<00:00, 20.49it/s, loss=0] Epoch: 286: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 286: 100% 213/213 [00:10<00:00, 20.41it/s, loss=0] Epoch: 287: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 287: 100% 213/213 [00:10<00:00, 20.40it/s, loss=0] Epoch: 288: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 288: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 289: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 289: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0] Epoch: 290: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 290: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 291: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 291: 100% 213/213 [00:10<00:00, 20.46it/s, loss=0] Epoch: 292: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 292: 100% 213/213 [00:10<00:00, 20.43it/s, loss=0] Epoch: 293: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 293: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 294: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 294: 100% 213/213 [00:10<00:00, 20.45it/s, loss=0] Epoch: 295: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 295: 100% 213/213 [00:10<00:00, 20.45it/s, loss=0] Epoch: 296: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 296: 100% 213/213 [00:10<00:00, 20.44it/s, loss=0] Epoch: 297: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 297: 100% 213/213 [00:10<00:00, 20.39it/s, loss=0] Epoch: 298: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 298: 100% 213/213 [00:10<00:00, 20.48it/s, loss=0] Epoch: 299: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 299: 100% 213/213 [00:10<00:00, 20.47it/s, loss=0] Epoch: 100% 300/300 [52:20<00:00, 10.47s/it] INFO:root:Training is finished ... Iteration: 100% 24/24 [00:10<00:00, 2.26it/s, acc=1] Iteration: 100% 165/165 [22:22<00:00, 8.14s/it, acc=0.85] obj_pp_to_subj_pp: 13.5 cp_recursion: 50.0 pp_recursion: 21.4 subj_to_obj_proper: 90.6 prim_to_obj_proper: 90.1 prim_to_subj_proper: 100.0 LEX: 94.69333333333333 OVERALL: 85.04761904761905 INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 4344077 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.99 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:10<00:00, 20.08it/s, loss=5.87] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:10<00:00, 20.27it/s, loss=4.55] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:10<00:00, 20.24it/s, loss=3.48] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:10<00:00, 20.18it/s, loss=2.44] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:10<00:00, 20.23it/s, loss=1.91] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:10<00:00, 20.28it/s, loss=1.58] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:10<00:00, 20.24it/s, loss=1.32] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:10<00:00, 20.22it/s, loss=1.14] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:10<00:00, 20.24it/s, loss=1.01] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0.89] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0.79] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0.7] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0.61] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0.53] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0.46] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0.39] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0.33] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0.28] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0.23] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0.2] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0.16] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0.14] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0.11] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0.09] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0.08] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0.07] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0.06] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0.05] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0.04] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0.04] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0.04] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0.03] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0.03] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0.03] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0.02] Epoch: 35: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 35: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0.02] Epoch: 36: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 36: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0.02] Epoch: 37: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 37: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0.02] Epoch: 38: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 38: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0.01] Epoch: 39: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 39: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0.01] Epoch: 40: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 40: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0.01] Epoch: 41: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 41: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0.01] Epoch: 42: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 42: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0.01] Epoch: 43: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 43: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0.01] Epoch: 44: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 44: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0.01] Epoch: 45: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 45: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0.01] Epoch: 46: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 46: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0.01] Epoch: 47: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 47: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0.01] Epoch: 48: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 48: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0.01] Epoch: 49: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 49: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0.01] Epoch: 50: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 50: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0.01] Epoch: 51: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 51: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0.01] Epoch: 52: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 52: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0.01] Epoch: 53: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 53: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 54: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 54: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 55: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 55: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0.01] Epoch: 56: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 56: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0.01] Epoch: 57: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 57: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 58: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 58: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 59: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 59: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0.01] Epoch: 60: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 60: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0.01] Epoch: 61: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 61: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 62: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 62: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 63: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 63: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 64: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 64: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 65: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 65: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 66: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 66: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 67: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 67: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 68: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 68: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 69: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 69: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 70: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 70: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 71: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 71: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 72: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 72: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 73: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 73: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 74: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 74: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 75: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 75: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 76: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 76: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0.01] Epoch: 77: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 77: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 78: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 78: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 79: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 79: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 80: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 80: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 81: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 81: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 82: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 82: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 83: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 83: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 84: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 84: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 85: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 85: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 86: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 86: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 87: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 87: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 88: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 88: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 89: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 89: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 90: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 90: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 91: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 91: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 92: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 92: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 93: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 93: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 94: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 94: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 95: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 95: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 96: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 96: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 97: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 97: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 98: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 98: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 99: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 99: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 100: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 100: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 101: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 101: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 102: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 102: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 103: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 103: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 104: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 104: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 105: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 105: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 106: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 106: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 107: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 107: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 108: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 108: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 109: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 109: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 110: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 110: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 111: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 111: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 112: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 112: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 113: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 113: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 114: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 114: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 115: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 115: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 116: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 116: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 117: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 117: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 118: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 118: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 119: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 119: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 120: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 120: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 121: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 121: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 122: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 122: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 123: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 123: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 124: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 124: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 125: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 125: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 126: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 126: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 127: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 127: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 128: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 128: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 129: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 129: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 130: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 130: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 131: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 131: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 132: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 132: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 133: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 133: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 134: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 134: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 135: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 135: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 136: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 136: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 137: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 137: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 138: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 138: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 139: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 139: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 140: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 140: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 141: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 141: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 142: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 142: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 143: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 143: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 144: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 144: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 145: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 145: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 146: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 146: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 147: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 147: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 148: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 148: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 149: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 149: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 150: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 150: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 151: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 151: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 152: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 152: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 153: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 153: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 154: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 154: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 155: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 155: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 156: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 156: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 157: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 157: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 158: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 158: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 159: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 159: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 160: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 160: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 161: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 161: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 162: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 162: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 163: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 163: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 164: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 164: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 165: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 165: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 166: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 166: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 167: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 167: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 168: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 168: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 169: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 169: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 170: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 170: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 171: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 171: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 172: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 172: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 173: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 173: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 174: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 174: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 175: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 175: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 176: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 176: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 177: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 177: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 178: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 178: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 179: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 179: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 180: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 180: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 181: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 181: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 182: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 182: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 183: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 183: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 184: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 184: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 185: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 185: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 186: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 186: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 187: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 187: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 188: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 188: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 189: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 189: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 190: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 190: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 191: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 191: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 192: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 192: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 193: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 193: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 194: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 194: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 195: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 195: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 196: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 196: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 197: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 197: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 198: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 198: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 199: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 199: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 200: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 200: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 201: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 201: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 202: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 202: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 203: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 203: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 204: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 204: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 205: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 205: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 206: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 206: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 207: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 207: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 208: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 208: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 209: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 209: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 210: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 210: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 211: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 211: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 212: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 212: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 213: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 213: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 214: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 214: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 215: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 215: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 216: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 216: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 217: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 217: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 218: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 218: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 219: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 219: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 220: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 220: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 221: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 221: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 222: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 222: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 223: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 223: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 224: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 224: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 225: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 225: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 226: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 226: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 227: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 227: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 228: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 228: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 229: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 229: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 230: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 230: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 231: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 231: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 232: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 232: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 233: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 233: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 234: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 234: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 235: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 235: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 236: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 236: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 237: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 237: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 238: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 238: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 239: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 239: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 240: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 240: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 241: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 241: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 242: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 242: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 243: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 243: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 244: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 244: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 245: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 245: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 246: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 246: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 247: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 247: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 248: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 248: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 249: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 249: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 250: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 250: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 251: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 251: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 252: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 252: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 253: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 253: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 254: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 254: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 255: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 255: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 256: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 256: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 257: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 257: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 258: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 258: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 259: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 259: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 260: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 260: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 261: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 261: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 262: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 262: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 263: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 263: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 264: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 264: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 265: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 265: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 266: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 266: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 267: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 267: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 268: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 268: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 269: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 269: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 270: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 270: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 271: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 271: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 272: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 272: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 273: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 273: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 274: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 274: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 275: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 275: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 276: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 276: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 277: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 277: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 278: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 278: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 279: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 279: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 280: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 280: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 281: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 281: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 282: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 282: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 283: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 283: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 284: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 284: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 285: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 285: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 286: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 286: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 287: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 287: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 288: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 288: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 289: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 289: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 290: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 290: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 291: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 291: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 292: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 292: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 293: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 293: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 294: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 294: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 295: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 295: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 296: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 296: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 297: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 297: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 298: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 298: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 299: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 299: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 100% 300/300 [52:56<00:00, 10.59s/it] INFO:root:Training is finished ... Iteration: 100% 24/24 [00:10<00:00, 2.25it/s, acc=1] Iteration: 100% 165/165 [11:42<00:00, 4.26s/it, acc=0.858] obj_pp_to_subj_pp: 17.8 cp_recursion: 51.3 pp_recursion: 48.2 subj_to_obj_proper: 94.2 prim_to_obj_proper: 91.1 prim_to_subj_proper: 100.0 LEX: 93.24 OVERALL: 85.77142857142857
more seeds
In [ ]:
!python run_cogs.py --model_name ende_transformer --gpu 1 --train_batch_size 128 --eval_batch_size 128 --lr 0.0001 --data_path ./recogs_positional_index --output_dir ./results_recogs_positional_index --lfs cogs --do_train --do_test --do_gen --max_seq_len 512 --output_json --epochs 300 --seeds "43;67;78;89;100" # paper's seeds +1
INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 4344077 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.43 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]We strongly recommend passing in an `attention_mask` since your input_ids may be padded. See https://huggingface.co/docs/transformers/troubleshooting#incorrect-output-when-padding-tokens-arent-masked. /content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:12<00:00, 17.33it/s, loss=6] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:10<00:00, 20.07it/s, loss=4.59] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:10<00:00, 20.08it/s, loss=3.54] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:10<00:00, 20.10it/s, loss=2.48] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:10<00:00, 20.07it/s, loss=1.92] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:10<00:00, 20.03it/s, loss=1.55] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:10<00:00, 20.01it/s, loss=1.26] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:10<00:00, 20.05it/s, loss=1.08] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.96] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0.87] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0.77] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0.68] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.59] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.51] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.45] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.38] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0.32] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0.27] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0.22] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.19] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.15] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.12] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0.1] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.08] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.06] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.06] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0.04] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.03] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0.03] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.03] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0.03] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.03] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.02] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.02] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0.02] Epoch: 35: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 35: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0.01] Epoch: 36: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 36: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0.01] Epoch: 37: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 37: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.01] Epoch: 38: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 38: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.01] Epoch: 39: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 39: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.01] Epoch: 40: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 40: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0.01] Epoch: 41: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 41: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.01] Epoch: 42: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 42: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.01] Epoch: 43: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 43: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.01] Epoch: 44: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 44: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.01] Epoch: 45: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 45: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0.01] Epoch: 46: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 46: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 47: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 47: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 48: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 48: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 49: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 49: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 50: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 50: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 51: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 51: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 52: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 52: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.01] Epoch: 53: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 53: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 54: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 54: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 55: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 55: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 56: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 56: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0.01] Epoch: 57: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 57: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 58: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 58: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.01] Epoch: 59: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 59: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 60: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 60: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 61: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 61: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 62: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 62: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 63: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 63: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 64: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 64: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 65: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 65: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 66: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 66: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 67: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 67: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 68: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 68: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 69: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 69: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 70: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 70: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 71: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 71: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 72: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 72: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 73: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 73: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 74: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 74: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 75: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 75: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 76: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 76: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 77: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 77: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 78: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 78: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 79: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 79: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 80: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 80: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 81: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 81: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 82: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 82: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 83: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 83: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 84: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 84: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 85: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 85: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 86: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 86: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 87: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 87: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 88: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 88: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 89: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 89: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 90: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 90: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 91: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 91: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 92: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 92: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 93: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 93: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 94: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 94: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 95: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 95: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 96: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 96: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 97: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 97: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 98: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 98: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 99: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 99: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 100: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 100: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 101: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 101: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 102: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 102: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 103: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 103: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 104: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 104: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 105: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 105: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 106: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 106: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 107: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 107: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0] Epoch: 108: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 108: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 109: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 109: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 110: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 110: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 111: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 111: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 112: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 112: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 113: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 113: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 114: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 114: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 115: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 115: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 116: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 116: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 117: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 117: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 118: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 118: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 119: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 119: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 120: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 120: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 121: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 121: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 122: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 122: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 123: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 123: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 124: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 124: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 125: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 125: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 126: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 126: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 127: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 127: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 128: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 128: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 129: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 129: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 130: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 130: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 131: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 131: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 132: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 132: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 133: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 133: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 134: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 134: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 135: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 135: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 136: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 136: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 137: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 137: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 138: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 138: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 139: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 139: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 140: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 140: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 141: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 141: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 142: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 142: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 143: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 143: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 144: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 144: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 145: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 145: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 146: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 146: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 147: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 147: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 148: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 148: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 149: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 149: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 150: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 150: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 151: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 151: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 152: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 152: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 153: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 153: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 154: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 154: 100% 213/213 [00:10<00:00, 19.95it/s, loss=0] Epoch: 155: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 155: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 156: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 156: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 157: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 157: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 158: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 158: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 159: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 159: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 160: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 160: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 161: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 161: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 162: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 162: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 163: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 163: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 164: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 164: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 165: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 165: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 166: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 166: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 167: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 167: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 168: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 168: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 169: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 169: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 170: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 170: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 171: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 171: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 172: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 172: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 173: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 173: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 174: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 174: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 175: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 175: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 176: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 176: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 177: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 177: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 178: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 178: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 179: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 179: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 180: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 180: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 181: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 181: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 182: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 182: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 183: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 183: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 184: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 184: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 185: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 185: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 186: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 186: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 187: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 187: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 188: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 188: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 189: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 189: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 190: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 190: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 191: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 191: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 192: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 192: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 193: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 193: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 194: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 194: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 195: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 195: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 196: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 196: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 197: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 197: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 198: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 198: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 199: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 199: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 200: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 200: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 201: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 201: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 202: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 202: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 203: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 203: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 204: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 204: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 205: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 205: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 206: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 206: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 207: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 207: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 208: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 208: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 209: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 209: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 210: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 210: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 211: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 211: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 212: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 212: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 213: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 213: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 214: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 214: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 215: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 215: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 216: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 216: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 217: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 217: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 218: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 218: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 219: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 219: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 220: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 220: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 221: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 221: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 222: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 222: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 223: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 223: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 224: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 224: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 225: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 225: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 226: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 226: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 227: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 227: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 228: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 228: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 229: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 229: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 230: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 230: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 231: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 231: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 232: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 232: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 233: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 233: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 234: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 234: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 235: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 235: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 236: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 236: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 237: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 237: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 238: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 238: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 239: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 239: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 240: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 240: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 241: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 241: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 242: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 242: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 243: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 243: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 244: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 244: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 245: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 245: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 246: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 246: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 247: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 247: 100% 213/213 [00:10<00:00, 19.95it/s, loss=0] Epoch: 248: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 248: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 249: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 249: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 250: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 250: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 251: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 251: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 252: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 252: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 253: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 253: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 254: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 254: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 255: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 255: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 256: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 256: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 257: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 257: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 258: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 258: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 259: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 259: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 260: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 260: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 261: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 261: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 262: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 262: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 263: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 263: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 264: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 264: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 265: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 265: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 266: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 266: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 267: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 267: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 268: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 268: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 269: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 269: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 270: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 270: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 271: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 271: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 272: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 272: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 273: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 273: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 274: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 274: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 275: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 275: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 276: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 276: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 277: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 277: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 278: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 278: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 279: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 279: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 280: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 280: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 281: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 281: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 282: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 282: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 283: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 283: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 284: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 284: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 285: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 285: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 286: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 286: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 287: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 287: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 288: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 288: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 289: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 289: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 290: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 290: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 291: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 291: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 292: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 292: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 293: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 293: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 294: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 294: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 295: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 295: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 296: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 296: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 297: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 297: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 298: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 298: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 299: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 299: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 100% 300/300 [53:25<00:00, 10.69s/it] INFO:root:Training is finished ... Iteration: 100% 24/24 [00:10<00:00, 2.28it/s, acc=1] Iteration: 100% 165/165 [17:57<00:00, 6.53s/it, acc=0.905] obj_pp_to_subj_pp: 20.2 cp_recursion: 52.0 pp_recursion: 61.8 subj_to_obj_proper: 88.4 prim_to_obj_proper: 83.0 prim_to_subj_proper: 100.0 LEX: 99.62666666666667 OVERALL: 90.46666666666667 INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 4344077 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.67 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:10<00:00, 19.86it/s, loss=6.01] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:10<00:00, 20.11it/s, loss=4.63] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:10<00:00, 20.03it/s, loss=3.63] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:10<00:00, 20.05it/s, loss=2.54] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:10<00:00, 20.11it/s, loss=1.94] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:10<00:00, 20.11it/s, loss=1.58] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:10<00:00, 20.04it/s, loss=1.32] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:10<00:00, 20.11it/s, loss=1.15] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:10<00:00, 20.10it/s, loss=1.03] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.92] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.83] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.74] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0.67] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0.6] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0.52] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0.46] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.38] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0.34] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.29] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0.26] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.22] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.2] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0.18] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0.17] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0.14] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.14] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0.12] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.1] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0.1] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0.1] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.1] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.08] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0.08] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.07] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.07] Epoch: 35: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 35: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0.06] Epoch: 36: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 36: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.06] Epoch: 37: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 37: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.06] Epoch: 38: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 38: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.05] Epoch: 39: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 39: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.05] Epoch: 40: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 40: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0.05] Epoch: 41: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 41: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.04] Epoch: 42: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 42: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.05] Epoch: 43: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 43: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.04] Epoch: 44: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 44: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.04] Epoch: 45: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 45: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.05] Epoch: 46: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 46: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0.04] Epoch: 47: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 47: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0.04] Epoch: 48: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 48: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.04] Epoch: 49: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 49: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.03] Epoch: 50: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 50: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.03] Epoch: 51: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 51: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0.03] Epoch: 52: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 52: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0.03] Epoch: 53: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 53: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0.03] Epoch: 54: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 54: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0.03] Epoch: 55: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 55: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0.03] Epoch: 56: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 56: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0.03] Epoch: 57: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 57: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0.02] Epoch: 58: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 58: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0.03] Epoch: 59: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 59: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0.02] Epoch: 60: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 60: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.03] Epoch: 61: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 61: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.02] Epoch: 62: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 62: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0.02] Epoch: 63: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 63: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0.02] Epoch: 64: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 64: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.02] Epoch: 65: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 65: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0.02] Epoch: 66: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 66: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.02] Epoch: 67: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 67: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0.02] Epoch: 68: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 68: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0.02] Epoch: 69: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 69: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0.02] Epoch: 70: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 70: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.01] Epoch: 71: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 71: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.01] Epoch: 72: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 72: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0.02] Epoch: 73: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 73: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.02] Epoch: 74: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 74: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.01] Epoch: 75: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 75: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0.01] Epoch: 76: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 76: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0.01] Epoch: 77: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 77: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.01] Epoch: 78: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 78: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0.01] Epoch: 79: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 79: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0.01] Epoch: 80: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 80: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.01] Epoch: 81: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 81: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0.01] Epoch: 82: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 82: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0.01] Epoch: 83: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 83: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0.01] Epoch: 84: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 84: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.01] Epoch: 85: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 85: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0.01] Epoch: 86: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 86: 100% 213/213 [00:10<00:00, 19.95it/s, loss=0.01] Epoch: 87: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 87: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0.01] Epoch: 88: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 88: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.01] Epoch: 89: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 89: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0.01] Epoch: 90: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 90: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0.01] Epoch: 91: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 91: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0.01] Epoch: 92: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 92: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0.01] Epoch: 93: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 93: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0.01] Epoch: 94: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 94: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0.01] Epoch: 95: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 95: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.01] Epoch: 96: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 96: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 97: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 97: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 98: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 98: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0.01] Epoch: 99: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 99: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0.01] Epoch: 100: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 100: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 101: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 101: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.01] Epoch: 102: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 102: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0.01] Epoch: 103: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 103: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0.01] Epoch: 104: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 104: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 105: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 105: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.01] Epoch: 106: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 106: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0.01] Epoch: 107: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 107: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 108: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 108: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 109: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 109: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 110: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 110: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 111: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 111: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 112: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 112: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 113: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 113: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 114: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 114: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 115: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 115: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 116: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 116: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 117: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 117: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 118: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 118: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 119: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 119: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 120: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 120: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 121: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 121: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 122: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 122: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 123: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 123: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.01] Epoch: 124: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 124: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0.01] Epoch: 125: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 125: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 126: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 126: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 127: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 127: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 128: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 128: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 129: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 129: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 130: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 130: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 131: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 131: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 132: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 132: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 133: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 133: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 134: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 134: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 135: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 135: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 136: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 136: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 137: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 137: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 138: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 138: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 139: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 139: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.01] Epoch: 140: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 140: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 141: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 141: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 142: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 142: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 143: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 143: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 144: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 144: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 145: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 145: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 146: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 146: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 147: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 147: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 148: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 148: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 149: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 149: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 150: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 150: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 151: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 151: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 152: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 152: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 153: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 153: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 154: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 154: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 155: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 155: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 156: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 156: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 157: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 157: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 158: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 158: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 159: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 159: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 160: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 160: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 161: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 161: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 162: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 162: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 163: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 163: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 164: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 164: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 165: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 165: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 166: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 166: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 167: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 167: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 168: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 168: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 169: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 169: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 170: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 170: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 171: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 171: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 172: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 172: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 173: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 173: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 174: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 174: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 175: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 175: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 176: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 176: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 177: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 177: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 178: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 178: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 179: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 179: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 180: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 180: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 181: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 181: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 182: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 182: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 183: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 183: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 184: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 184: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 185: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 185: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 186: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 186: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 187: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 187: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 188: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 188: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 189: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 189: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 190: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 190: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 191: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 191: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 192: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 192: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 193: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 193: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 194: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 194: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 195: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 195: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 196: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 196: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 197: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 197: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 198: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 198: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 199: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 199: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 200: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 200: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 201: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 201: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 202: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 202: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 203: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 203: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 204: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 204: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 205: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 205: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 206: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 206: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 207: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 207: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 208: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 208: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 209: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 209: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 210: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 210: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 211: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 211: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 212: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 212: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 213: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 213: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 214: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 214: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 215: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 215: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 216: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 216: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 217: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 217: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 218: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 218: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 219: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 219: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 220: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 220: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 221: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 221: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 222: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 222: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 223: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 223: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 224: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 224: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 225: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 225: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 226: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 226: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 227: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 227: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 228: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 228: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 229: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 229: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 230: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 230: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 231: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 231: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 232: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 232: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 233: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 233: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 234: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 234: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 235: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 235: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 236: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 236: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 237: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 237: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 238: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 238: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 239: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 239: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 240: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 240: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 241: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 241: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 242: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 242: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 243: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 243: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 244: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 244: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 245: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 245: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 246: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 246: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 247: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 247: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 248: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 248: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 249: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 249: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 250: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 250: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 251: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 251: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 252: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 252: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 253: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 253: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 254: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 254: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 255: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 255: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 256: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 256: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 257: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 257: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 258: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 258: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 259: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 259: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 260: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 260: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 261: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 261: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 262: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 262: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 263: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 263: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 264: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 264: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 265: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 265: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 266: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 266: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 267: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 267: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 268: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 268: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 269: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 269: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 270: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 270: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 271: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 271: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 272: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 272: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 273: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 273: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 274: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 274: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 275: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 275: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 276: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 276: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 277: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 277: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 278: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 278: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 279: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 279: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 280: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 280: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 281: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 281: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 282: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 282: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 283: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 283: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 284: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 284: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 285: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 285: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 286: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 286: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 287: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 287: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 288: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 288: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 289: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 289: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 290: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 290: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 291: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 291: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 292: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 292: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 293: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 293: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 294: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 294: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 295: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 295: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 296: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 296: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 297: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 297: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 298: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 298: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 299: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 299: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 100% 300/300 [53:21<00:00, 10.67s/it] INFO:root:Training is finished ... Iteration: 100% 24/24 [00:10<00:00, 2.27it/s, acc=1] Iteration: 100% 165/165 [23:03<00:00, 8.38s/it, acc=0.878] obj_pp_to_subj_pp: 17.5 cp_recursion: 52.8 pp_recursion: 32.0 subj_to_obj_proper: 82.8 prim_to_obj_proper: 63.4 prim_to_subj_proper: 100.0 LEX: 99.73333333333333 OVERALL: 87.83333333333333 INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 4344077 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.78 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:10<00:00, 19.94it/s, loss=5.96] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:10<00:00, 20.09it/s, loss=4.65] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:10<00:00, 20.16it/s, loss=3.57] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:10<00:00, 20.16it/s, loss=2.5] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:10<00:00, 20.13it/s, loss=1.94] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:10<00:00, 20.11it/s, loss=1.61] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:10<00:00, 20.02it/s, loss=1.35] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:10<00:00, 20.03it/s, loss=1.17] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:10<00:00, 20.13it/s, loss=1.03] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0.91] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0.82] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0.72] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0.63] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.55] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0.48] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0.42] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0.36] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0.32] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0.27] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0.24] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0.21] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0.17] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.16] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0.14] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.14] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0.12] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0.11] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0.09] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0.09] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0.07] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.07] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0.06] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0.06] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0.05] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 2% 5/213 [00:00<00:10, 19.36it/s, loss=0.03]
ran out of money, pickup from next seed
In [ ]:
!python run_cogs.py --model_name ende_transformer --gpu 1 --train_batch_size 128 --eval_batch_size 128 --lr 0.0001 --data_path ./recogs_positional_index --output_dir ./results_recogs_positional_index --lfs cogs --do_train --do_test --do_gen --max_seq_len 512 --output_json --epochs 300 --seeds "78;89;100" # paper's seeds +1
INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 4344077 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.78 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]We strongly recommend passing in an `attention_mask` since your input_ids may be padded. See https://huggingface.co/docs/transformers/troubleshooting#incorrect-output-when-padding-tokens-arent-masked. /content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:12<00:00, 17.60it/s, loss=5.96] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:10<00:00, 20.08it/s, loss=4.65] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:10<00:00, 20.15it/s, loss=3.57] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:10<00:00, 20.09it/s, loss=2.5] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:10<00:00, 20.25it/s, loss=1.94] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:10<00:00, 20.22it/s, loss=1.61] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:10<00:00, 20.25it/s, loss=1.35] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:10<00:00, 20.23it/s, loss=1.17] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:10<00:00, 20.26it/s, loss=1.03] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0.91] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0.82] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0.72] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0.63] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0.55] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0.48] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0.42] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0.36] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0.32] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0.27] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0.24] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0.21] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0.17] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0.16] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0.14] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0.14] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0.12] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0.11] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0.09] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0.09] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0.07] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0.07] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0.06] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0.06] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0.05] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0.05] Epoch: 35: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 35: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0.04] Epoch: 36: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 36: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0.04] Epoch: 37: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 37: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0.04] Epoch: 38: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 38: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0.03] Epoch: 39: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 39: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0.03] Epoch: 40: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 40: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0.02] Epoch: 41: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 41: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0.03] Epoch: 42: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 42: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0.02] Epoch: 43: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 43: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0.02] Epoch: 44: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 44: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0.02] Epoch: 45: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 45: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0.02] Epoch: 46: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 46: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0.02] Epoch: 47: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 47: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0.01] Epoch: 48: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 48: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0.01] Epoch: 49: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 49: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0.01] Epoch: 50: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 50: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0.01] Epoch: 51: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 51: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0.01] Epoch: 52: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 52: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0.02] Epoch: 53: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 53: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0.01] Epoch: 54: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 54: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0.01] Epoch: 55: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 55: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0.01] Epoch: 56: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 56: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0.01] Epoch: 57: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 57: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0.01] Epoch: 58: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 58: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0.01] Epoch: 59: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 59: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0.01] Epoch: 60: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 60: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0.01] Epoch: 61: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 61: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0.01] Epoch: 62: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 62: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0.01] Epoch: 63: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 63: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0.01] Epoch: 64: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 64: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0.01] Epoch: 65: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 65: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 66: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 66: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 67: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 67: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 68: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 68: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 69: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 69: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0.01] Epoch: 70: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 70: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0.01] Epoch: 71: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 71: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 72: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 72: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 73: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 73: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 74: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 74: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 75: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 75: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 76: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 76: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 77: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 77: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 78: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 78: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 79: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 79: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0.01] Epoch: 80: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 80: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0.01] Epoch: 81: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 81: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 82: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 82: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 83: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 83: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 84: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 84: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 85: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 85: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 86: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 86: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 87: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 87: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 88: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 88: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 89: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 89: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 90: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 90: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 91: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 91: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 92: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 92: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 93: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 93: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 94: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 94: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 95: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 95: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 96: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 96: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 97: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 97: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 98: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 98: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 99: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 99: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 100: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 100: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 101: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 101: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 102: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 102: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 103: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 103: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 104: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 104: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 105: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 105: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 106: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 106: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 107: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 107: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 108: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 108: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 109: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 109: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 110: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 110: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 111: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 111: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 112: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 112: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 113: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 113: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 114: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 114: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 115: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 115: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 116: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 116: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0.01] Epoch: 117: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 117: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 118: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 118: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 119: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 119: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0.01] Epoch: 120: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 120: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 121: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 121: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 122: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 122: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 123: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 123: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 124: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 124: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 125: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 125: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 126: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 126: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 127: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 127: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 128: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 128: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 129: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 129: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 130: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 130: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 131: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 131: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 132: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 132: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 133: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 133: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 134: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 134: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 135: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 135: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 136: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 136: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 137: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 137: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 138: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 138: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 139: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 139: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 140: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 140: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 141: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 141: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 142: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 142: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 143: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 143: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 144: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 144: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 145: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 145: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 146: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 146: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 147: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 147: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 148: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 148: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 149: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 149: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 150: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 150: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 151: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 151: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 152: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 152: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 153: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 153: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 154: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 154: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 155: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 155: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 156: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 156: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 157: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 157: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 158: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 158: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 159: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 159: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 160: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 160: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 161: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 161: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 162: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 162: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 163: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 163: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 164: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 164: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 165: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 165: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 166: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 166: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 167: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 167: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 168: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 168: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 169: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 169: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 170: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 170: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 171: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 171: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 172: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 172: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 173: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 173: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 174: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 174: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 175: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 175: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 176: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 176: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 177: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 177: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 178: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 178: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 179: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 179: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 180: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 180: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 181: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 181: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 182: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 182: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 183: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 183: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 184: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 184: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 185: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 185: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 186: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 186: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 187: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 187: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 188: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 188: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 189: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 189: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 190: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 190: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 191: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 191: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 192: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 192: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 193: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 193: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 194: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 194: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 195: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 195: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 196: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 196: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 197: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 197: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 198: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 198: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 199: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 199: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 200: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 200: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 201: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 201: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 202: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 202: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 203: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 203: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 204: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 204: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 205: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 205: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 206: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 206: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 207: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 207: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 208: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 208: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 209: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 209: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 210: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 210: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 211: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 211: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 212: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 212: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 213: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 213: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 214: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 214: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 215: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 215: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 216: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 216: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 217: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 217: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 218: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 218: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 219: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 219: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 220: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 220: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 221: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 221: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 222: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 222: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 223: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 223: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 224: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 224: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 225: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 225: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 226: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 226: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 227: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 227: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 228: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 228: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 229: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 229: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 230: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 230: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 231: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 231: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 232: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 232: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 233: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 233: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 234: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 234: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 235: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 235: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 236: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 236: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 237: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 237: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 238: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 238: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 239: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 239: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 240: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 240: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 241: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 241: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 242: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 242: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 243: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 243: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 244: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 244: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 245: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 245: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 246: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 246: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 247: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 247: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 248: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 248: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 249: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 249: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 250: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 250: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 251: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 251: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 252: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 252: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 253: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 253: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 254: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 254: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 255: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 255: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 256: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 256: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 257: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 257: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 258: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 258: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 259: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 259: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 260: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 260: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 261: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 261: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 262: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 262: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 263: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 263: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 264: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 264: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 265: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 265: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 266: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 266: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 267: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 267: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 268: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 268: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 269: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 269: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 270: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 270: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 271: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 271: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 272: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 272: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 273: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 273: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 274: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 274: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 275: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 275: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 276: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 276: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 277: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 277: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 278: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 278: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 279: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 279: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 280: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 280: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 281: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 281: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 282: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 282: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 283: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 283: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 284: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 284: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 285: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 285: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 286: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 286: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 287: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 287: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 288: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 288: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 289: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 289: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 290: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 290: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 291: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 291: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 292: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 292: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 293: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 293: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 294: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 294: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 295: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 295: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 296: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 296: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 297: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 297: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 298: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 298: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 299: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 299: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 100% 300/300 [52:58<00:00, 10.60s/it] INFO:root:Training is finished ... Iteration: 100% 24/24 [00:10<00:00, 2.31it/s, acc=1] Iteration: 100% 165/165 [17:26<00:00, 6.34s/it, acc=0.875] obj_pp_to_subj_pp: 20.0 cp_recursion: 49.3 pp_recursion: 50.1 subj_to_obj_proper: 94.7 prim_to_obj_proper: 88.8 prim_to_subj_proper: 100.0 LEX: 95.64666666666668 OVERALL: 87.5047619047619 INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 4344077 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.89 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:10<00:00, 20.16it/s, loss=6.03] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:10<00:00, 20.27it/s, loss=4.65] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:10<00:00, 20.26it/s, loss=3.6] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:10<00:00, 20.21it/s, loss=2.53] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:10<00:00, 20.24it/s, loss=1.97] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:10<00:00, 20.25it/s, loss=1.61] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:10<00:00, 20.24it/s, loss=1.34] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:10<00:00, 20.25it/s, loss=1.16] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:10<00:00, 20.28it/s, loss=1.03] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0.93] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0.84] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0.74] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0.66] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0.58] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0.5] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0.42] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0.37] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0.31] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0.26] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0.23] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0.19] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0.16] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0.15] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0.12] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0.11] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0.1] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0.09] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0.09] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0.07] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0.06] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0.06] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0.05] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0.06] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0.05] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0.05] Epoch: 35: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 35: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0.04] Epoch: 36: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 36: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0.04] Epoch: 37: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 37: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0.04] Epoch: 38: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 38: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.03] Epoch: 39: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 39: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0.03] Epoch: 40: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 40: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0.03] Epoch: 41: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 41: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0.03] Epoch: 42: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 42: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0.03] Epoch: 43: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 43: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0.02] Epoch: 44: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 44: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0.02] Epoch: 45: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 45: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0.02] Epoch: 46: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 46: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0.03] Epoch: 47: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 47: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0.02] Epoch: 48: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 48: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0.02] Epoch: 49: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 49: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0.02] Epoch: 50: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 50: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0.02] Epoch: 51: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 51: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0.02] Epoch: 52: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 52: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0.02] Epoch: 53: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 53: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0.01] Epoch: 54: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 54: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.02] Epoch: 55: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 55: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0.01] Epoch: 56: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 56: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0.01] Epoch: 57: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 57: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0.01] Epoch: 58: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 58: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0.01] Epoch: 59: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 59: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0.02] Epoch: 60: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 60: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0.01] Epoch: 61: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 61: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0.01] Epoch: 62: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 62: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0.01] Epoch: 63: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 63: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0.01] Epoch: 64: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 64: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0.01] Epoch: 65: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 65: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0.01] Epoch: 66: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 66: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0.01] Epoch: 67: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 67: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0.01] Epoch: 68: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 68: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0.01] Epoch: 69: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 69: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0.01] Epoch: 70: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 70: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0.01] Epoch: 71: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 71: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0.01] Epoch: 72: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 72: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0.01] Epoch: 73: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 73: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0.01] Epoch: 74: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 74: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0.01] Epoch: 75: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 75: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0.01] Epoch: 76: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 76: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0.01] Epoch: 77: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 77: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0.01] Epoch: 78: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 78: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0.01] Epoch: 79: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 79: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 80: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 80: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 81: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 81: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 82: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 82: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0.01] Epoch: 83: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 83: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0.01] Epoch: 84: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 84: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0.01] Epoch: 85: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 85: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 86: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 86: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0.01] Epoch: 87: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 87: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0.01] Epoch: 88: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 88: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0.01] Epoch: 89: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 89: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0.01] Epoch: 90: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 90: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0.01] Epoch: 91: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 91: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 92: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 92: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0.01] Epoch: 93: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 93: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 94: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 94: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 95: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 95: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 96: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 96: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 97: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 97: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 98: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 98: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0.01] Epoch: 99: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 99: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 100: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 100: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0.01] Epoch: 101: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 101: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 102: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 102: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 103: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 103: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0.01] Epoch: 104: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 104: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 105: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 105: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 106: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 106: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 107: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 107: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 108: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 108: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 109: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 109: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0.01] Epoch: 110: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 110: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 111: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 111: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 112: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 112: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 113: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 113: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 114: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 114: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 115: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 115: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 116: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 116: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 117: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 117: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 118: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 118: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 119: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 119: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 120: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 120: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0.01] Epoch: 121: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 121: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 122: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 122: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 123: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 123: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 124: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 124: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 125: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 125: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 126: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 126: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 127: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 127: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 128: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 128: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 129: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 129: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 130: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 130: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 131: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 131: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 132: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 132: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 133: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 133: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0.01] Epoch: 134: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 134: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 135: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 135: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 136: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 136: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 137: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 137: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 138: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 138: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 139: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 139: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 140: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 140: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 141: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 141: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 142: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 142: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 143: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 143: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 144: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 144: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 145: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 145: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 146: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 146: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 147: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 147: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 148: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 148: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 149: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 149: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 150: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 150: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 151: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 151: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 152: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 152: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 153: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 153: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 154: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 154: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 155: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 155: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 156: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 156: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 157: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 157: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 158: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 158: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 159: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 159: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 160: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 160: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 161: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 161: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 162: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 162: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 163: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 163: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 164: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 164: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 165: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 165: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 166: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 166: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 167: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 167: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 168: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 168: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 169: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 169: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 170: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 170: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 171: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 171: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 172: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 172: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 173: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 173: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 174: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 174: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 175: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 175: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 176: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 176: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 177: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 177: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 178: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 178: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 179: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 179: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 180: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 180: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 181: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 181: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 182: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 182: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 183: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 183: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 184: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 184: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 185: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 185: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 186: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 186: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 187: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 187: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 188: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 188: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 189: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 189: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 190: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 190: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 191: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 191: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 192: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 192: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 193: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 193: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 194: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 194: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 195: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 195: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 196: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 196: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 197: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 197: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 198: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 198: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 199: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 199: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 200: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 200: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 201: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 201: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 202: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 202: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 203: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 203: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 204: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 204: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 205: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 205: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 206: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 206: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 207: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 207: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 208: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 208: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 209: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 209: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 210: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 210: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 211: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 211: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 212: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 212: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 213: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 213: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 214: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 214: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 215: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 215: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 216: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 216: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 217: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 217: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 218: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 218: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 219: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 219: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 220: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 220: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 221: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 221: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 222: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 222: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 223: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 223: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 224: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 224: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 225: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 225: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 226: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 226: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 227: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 227: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 228: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 228: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 229: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 229: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 230: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 230: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 231: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 231: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 232: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 232: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 233: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 233: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 234: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 234: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 235: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 235: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 236: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 236: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 237: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 237: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 238: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 238: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 239: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 239: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 240: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 240: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 241: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 241: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 242: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 242: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 243: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 243: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 244: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 244: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 245: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 245: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 246: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 246: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 247: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 247: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 248: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 248: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 249: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 249: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 250: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 250: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 251: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 251: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 252: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 252: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0] Epoch: 253: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 253: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 254: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 254: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0] Epoch: 255: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 255: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 256: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 256: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 257: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 257: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 258: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 258: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 259: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 259: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 260: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 260: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 261: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 261: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 262: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 262: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 263: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 263: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0] Epoch: 264: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 264: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 265: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 265: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 266: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 266: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0] Epoch: 267: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 267: 100% 213/213 [00:10<00:00, 20.30it/s, loss=0] Epoch: 268: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 268: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 269: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 269: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0] Epoch: 270: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 270: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 271: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 271: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 272: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 272: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 273: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 273: 100% 213/213 [00:10<00:00, 20.32it/s, loss=0] Epoch: 274: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 274: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 275: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 275: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 276: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 276: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 277: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 277: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 278: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 278: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 279: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 279: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 280: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 280: 100% 213/213 [00:10<00:00, 20.27it/s, loss=0] Epoch: 281: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 281: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 282: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 282: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 283: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 283: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 284: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 284: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 285: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 285: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 286: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 286: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 287: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 287: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 288: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 288: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 289: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 289: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 290: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 290: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 291: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 291: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 292: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 292: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 293: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 293: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 294: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 294: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 295: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 295: 100% 213/213 [00:10<00:00, 20.28it/s, loss=0] Epoch: 296: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 296: 100% 213/213 [00:10<00:00, 20.31it/s, loss=0] Epoch: 297: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 297: 100% 213/213 [00:10<00:00, 20.29it/s, loss=0] Epoch: 298: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 298: 100% 213/213 [00:10<00:00, 20.33it/s, loss=0] Epoch: 299: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 299: 100% 213/213 [00:10<00:00, 20.26it/s, loss=0] Epoch: 100% 300/300 [52:56<00:00, 10.59s/it] INFO:root:Training is finished ... Iteration: 100% 24/24 [00:10<00:00, 2.24it/s, acc=1] Iteration: 100% 165/165 [17:08<00:00, 6.23s/it, acc=0.897] obj_pp_to_subj_pp: 20.1 cp_recursion: 53.5 pp_recursion: 32.0 subj_to_obj_proper: 87.5 prim_to_obj_proper: 97.0 prim_to_subj_proper: 100.0 LEX: 99.52 OVERALL: 89.66190476190476 INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 4344077 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.100 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:10<00:00, 20.06it/s, loss=6.01] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:10<00:00, 20.22it/s, loss=4.57] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:10<00:00, 20.23it/s, loss=3.56] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:10<00:00, 20.15it/s, loss=2.5] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:10<00:00, 20.20it/s, loss=1.92] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:10<00:00, 20.11it/s, loss=1.58] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:10<00:00, 20.22it/s, loss=1.32] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:10<00:00, 20.23it/s, loss=1.15] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:10<00:00, 20.23it/s, loss=1.02] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0.92] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0.82] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0.72] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0.64] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0.56] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.49] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0.43] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0.37] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.31] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0.26] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0.22] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0.17] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0.15] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0.12] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0.1] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0.09] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0.07] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0.06] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0.06] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.05] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0.05] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0.04] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0.06] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0.04] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0.03] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.04] Epoch: 35: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 35: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0.02] Epoch: 36: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 36: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0.03] Epoch: 37: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 37: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0.02] Epoch: 38: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 38: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0.01] Epoch: 39: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 39: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0.02] Epoch: 40: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 40: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.02] Epoch: 41: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 41: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0.01] Epoch: 42: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 42: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0.01] Epoch: 43: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 43: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0.02] Epoch: 44: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 44: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0.01] Epoch: 45: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 45: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0.01] Epoch: 46: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 46: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.01] Epoch: 47: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 47: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0.01] Epoch: 48: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 48: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0.01] Epoch: 49: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 49: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0.01] Epoch: 50: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 50: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0.01] Epoch: 51: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 51: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0.01] Epoch: 52: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 52: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0.01] Epoch: 53: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 53: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0.01] Epoch: 54: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 54: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 55: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 55: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0.01] Epoch: 56: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 56: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.01] Epoch: 57: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 57: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0.01] Epoch: 58: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 58: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0.01] Epoch: 59: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 59: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 60: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 60: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0.01] Epoch: 61: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 61: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0] Epoch: 62: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 62: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 63: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 63: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0] Epoch: 64: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 64: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0.01] Epoch: 65: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 65: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 66: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 66: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0.01] Epoch: 67: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 67: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0.01] Epoch: 68: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 68: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 69: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 69: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0] Epoch: 70: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 70: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 71: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 71: 100% 213/213 [00:10<00:00, 19.95it/s, loss=0] Epoch: 72: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 72: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 73: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 73: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 74: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 74: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 75: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 75: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 76: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 76: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 77: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 77: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 78: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 78: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 79: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 79: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 80: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 80: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 81: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 81: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 82: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 82: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 83: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 83: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 84: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 84: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 85: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 85: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 86: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 86: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 87: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 87: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0.01] Epoch: 88: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 88: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 89: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 89: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 90: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 90: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 91: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 91: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 92: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 92: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 93: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 93: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 94: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 94: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 95: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 95: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 96: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 96: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 97: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 97: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 98: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 98: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 99: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 99: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 100: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 100: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 101: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 101: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 102: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 102: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 103: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 103: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 104: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 104: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 105: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 105: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 106: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 106: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 107: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 107: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 108: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 108: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 109: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 109: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 110: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 110: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 111: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 111: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 112: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 112: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 113: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 113: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 114: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 114: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 115: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 115: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 116: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 116: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 117: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 117: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 118: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 118: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 119: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 119: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 120: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 120: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 121: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 121: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 122: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 122: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 123: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 123: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 124: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 124: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 125: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 125: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 126: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 126: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 127: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 127: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 128: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 128: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 129: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 129: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 130: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 130: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 131: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 131: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 132: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 132: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 133: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 133: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 134: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 134: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 135: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 135: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 136: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 136: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 137: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 137: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 138: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 138: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 139: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 139: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 140: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 140: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 141: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 141: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 142: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 142: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 143: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 143: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 144: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 144: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 145: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 145: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 146: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 146: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 147: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 147: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 148: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 148: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 149: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 149: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 150: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 150: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 151: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 151: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 152: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 152: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 153: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 153: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 154: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 154: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 155: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 155: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 156: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 156: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 157: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 157: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 158: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 158: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 159: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 159: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 160: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 160: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 161: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 161: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 162: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 162: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 163: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 163: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 164: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 164: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 165: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 165: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 166: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 166: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 167: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 167: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 168: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 168: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 169: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 169: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 170: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 170: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 171: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 171: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 172: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 172: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 173: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 173: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 174: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 174: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 175: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 175: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 176: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 176: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 177: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 177: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 178: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 178: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 179: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 179: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 180: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 180: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 181: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 181: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 182: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 182: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 183: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 183: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 184: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 184: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 185: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 185: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 186: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 186: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 187: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 187: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 188: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 188: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 189: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 189: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 190: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 190: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 191: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 191: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 192: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 192: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 193: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 193: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 194: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 194: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 195: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 195: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 196: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 196: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 197: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 197: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 198: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 198: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 199: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 199: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 200: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 200: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 201: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 201: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 202: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 202: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 203: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 203: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 204: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 204: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 205: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 205: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 206: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 206: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 207: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 207: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 208: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 208: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 209: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 209: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 210: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 210: 100% 213/213 [00:10<00:00, 19.95it/s, loss=0] Epoch: 211: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 211: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 212: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 212: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 213: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 213: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 214: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 214: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 215: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 215: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 216: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 216: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 217: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 217: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 218: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 218: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 219: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 219: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 220: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 220: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 221: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 221: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 222: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 222: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 223: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 223: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 224: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 224: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 225: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 225: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 226: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 226: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 227: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 227: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 228: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 228: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 229: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 229: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 230: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 230: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 231: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 231: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 232: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 232: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 233: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 233: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 234: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 234: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 235: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 235: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 236: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 236: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 237: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 237: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 238: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 238: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 239: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 239: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 240: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 240: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 241: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 241: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 242: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 242: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 243: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 243: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 244: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 244: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 245: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 245: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 246: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 246: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 247: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 247: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 248: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 248: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 249: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 249: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 250: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 250: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 251: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 251: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 252: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 252: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 253: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 253: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 254: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 254: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 255: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 255: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 256: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 256: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 257: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 257: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 258: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 258: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 259: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 259: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 260: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 260: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 261: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 261: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 262: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 262: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 263: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 263: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 264: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 264: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 265: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 265: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 266: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 266: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 267: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 267: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 268: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 268: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 269: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 269: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 270: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 270: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 271: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 271: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 272: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 272: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 273: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 273: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 274: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 274: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 275: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 275: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 276: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 276: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 277: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 277: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 278: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 278: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 279: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 279: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 280: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 280: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 281: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 281: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 282: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 282: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 283: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 283: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 284: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 284: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 285: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 285: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 286: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 286: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 287: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 287: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 288: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 288: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 289: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 289: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 290: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 290: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 291: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 291: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 292: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 292: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 293: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 293: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 294: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 294: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 295: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 295: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 296: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 296: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 297: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 297: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 298: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 298: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 299: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 299: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 100% 300/300 [53:12<00:00, 10.64s/it] INFO:root:Training is finished ... Iteration: 100% 24/24 [00:10<00:00, 2.26it/s, acc=1] Iteration: 100% 165/165 [18:57<00:00, 6.90s/it, acc=0.884] obj_pp_to_subj_pp: 15.8 cp_recursion: 53.2 pp_recursion: 30.5 subj_to_obj_proper: 97.4 prim_to_obj_proper: 81.2 prim_to_subj_proper: 99.9 LEX: 98.58666666666667 OVERALL: 88.41904761904762
more seeds
In [ ]:
!python run_cogs.py --model_name ende_transformer --gpu 1 --train_batch_size 128 --eval_batch_size 128 --lr 0.0001 --data_path ./recogs_positional_index --output_dir ./results_recogs_positional_index --lfs cogs --do_train --do_test --do_gen --max_seq_len 512 --output_json --epochs 300 --seeds "44;68;79;90;101" # paper's seeds + 2, not sure how they chose them
INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 4344077 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.44 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]We strongly recommend passing in an `attention_mask` since your input_ids may be padded. See https://huggingface.co/docs/transformers/troubleshooting#incorrect-output-when-padding-tokens-arent-masked. /content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:12<00:00, 17.61it/s, loss=5.93] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:10<00:00, 20.06it/s, loss=4.56] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:10<00:00, 20.07it/s, loss=3.55] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:10<00:00, 19.99it/s, loss=2.52] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:10<00:00, 20.06it/s, loss=1.92] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:10<00:00, 20.05it/s, loss=1.58] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:10<00:00, 20.08it/s, loss=1.33] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:10<00:00, 20.04it/s, loss=1.15] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:10<00:00, 20.05it/s, loss=1.02] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0.93] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0.85] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0.77] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0.69] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0.62] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0.55] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0.49] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0.42] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0.35] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.29] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0.25] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0.22] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.18] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0.16] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0.13] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0.11] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0.1] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0.09] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.08] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.07] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0.07] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.07] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0.06] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0.05] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0.05] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.04] Epoch: 35: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 35: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0.04] Epoch: 36: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 36: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.04] Epoch: 37: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 37: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0.04] Epoch: 38: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 38: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0.03] Epoch: 39: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 39: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0.03] Epoch: 40: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 40: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0.02] Epoch: 41: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 41: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0.02] Epoch: 42: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 42: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0.04] Epoch: 43: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 43: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0.02] Epoch: 44: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 44: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0.02] Epoch: 45: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 45: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0.02] Epoch: 46: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 46: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.02] Epoch: 47: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 47: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0.02] Epoch: 48: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 48: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0.01] Epoch: 49: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 49: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0.02] Epoch: 50: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 50: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0.02] Epoch: 51: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 51: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0.02] Epoch: 52: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 52: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0.01] Epoch: 53: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 53: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0.01] Epoch: 54: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 54: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0.01] Epoch: 55: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 55: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0.01] Epoch: 56: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 56: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.01] Epoch: 57: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 57: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0.01] Epoch: 58: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 58: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.02] Epoch: 59: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 59: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0.01] Epoch: 60: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 60: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0.01] Epoch: 61: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 61: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0.01] Epoch: 62: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 62: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0.01] Epoch: 63: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 63: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0.01] Epoch: 64: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 64: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0.01] Epoch: 65: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 65: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0.01] Epoch: 66: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 66: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0.01] Epoch: 67: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 67: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.01] Epoch: 68: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 68: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0.01] Epoch: 69: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 69: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.01] Epoch: 70: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 70: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.01] Epoch: 71: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 71: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0.01] Epoch: 72: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 72: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0.01] Epoch: 73: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 73: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.01] Epoch: 74: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 74: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.01] Epoch: 75: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 75: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.01] Epoch: 76: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 76: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 77: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 77: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.01] Epoch: 78: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 78: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 79: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 79: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0.01] Epoch: 80: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 80: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0.01] Epoch: 81: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 81: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0.01] Epoch: 82: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 82: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0.01] Epoch: 83: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 83: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 84: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 84: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0.01] Epoch: 85: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 85: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 86: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 86: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 87: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 87: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 88: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 88: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.01] Epoch: 89: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 89: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 90: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 90: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 91: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 91: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.01] Epoch: 92: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 92: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 93: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 93: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 94: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 94: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 95: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 95: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 96: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 96: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 97: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 97: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0.01] Epoch: 98: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 98: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0.01] Epoch: 99: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 99: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 100: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 100: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 101: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 101: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 102: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 102: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 103: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 103: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 104: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 104: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 105: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 105: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 106: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 106: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 107: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 107: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 108: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 108: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 109: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 109: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0.01] Epoch: 110: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 110: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 111: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 111: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 112: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 112: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 113: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 113: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 114: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 114: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 115: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 115: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 116: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 116: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 117: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 117: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 118: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 118: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 119: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 119: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 120: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 120: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 121: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 121: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 122: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 122: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 123: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 123: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 124: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 124: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 125: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 125: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 126: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 126: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 127: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 127: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 128: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 128: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 129: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 129: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 130: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 130: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 131: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 131: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 132: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 132: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 133: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 133: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 134: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 134: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 135: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 135: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 136: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 136: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 137: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 137: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 138: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 138: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 139: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 139: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 140: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 140: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 141: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 141: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 142: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 142: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 143: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 143: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 144: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 144: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 145: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 145: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 146: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 146: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 147: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 147: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 148: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 148: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 149: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 149: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 150: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 150: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 151: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 151: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 152: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 152: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 153: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 153: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 154: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 154: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 155: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 155: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 156: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 156: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 157: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 157: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 158: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 158: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 159: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 159: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 160: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 160: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 161: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 161: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 162: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 162: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 163: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 163: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 164: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 164: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 165: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 165: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 166: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 166: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 167: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 167: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 168: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 168: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 169: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 169: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 170: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 170: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 171: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 171: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 172: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 172: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 173: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 173: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 174: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 174: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 175: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 175: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 176: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 176: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 177: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 177: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 178: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 178: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 179: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 179: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 180: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 180: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 181: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 181: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 182: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 182: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 183: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 183: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 184: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 184: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 185: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 185: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 186: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 186: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 187: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 187: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 188: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 188: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 189: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 189: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 190: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 190: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 191: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 191: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 192: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 192: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 193: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 193: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 194: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 194: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 195: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 195: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 196: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 196: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 197: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 197: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 198: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 198: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 199: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 199: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 200: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 200: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 201: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 201: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 202: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 202: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 203: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 203: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 204: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 204: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 205: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 205: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 206: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 206: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 207: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 207: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 208: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 208: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 209: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 209: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 210: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 210: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 211: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 211: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 212: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 212: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 213: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 213: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 214: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 214: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 215: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 215: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 216: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 216: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 217: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 217: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 218: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 218: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 219: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 219: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 220: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 220: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 221: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 221: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 222: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 222: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 223: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 223: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 224: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 224: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 225: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 225: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 226: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 226: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 227: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 227: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 228: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 228: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 229: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 229: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 230: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 230: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 231: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 231: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 232: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 232: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 233: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 233: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 234: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 234: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 235: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 235: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 236: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 236: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 237: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 237: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 238: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 238: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 239: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 239: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 240: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 240: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 241: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 241: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 242: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 242: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 243: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 243: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 244: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 244: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 245: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 245: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 246: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 246: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 247: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 247: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 248: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 248: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 249: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 249: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 250: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 250: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 251: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 251: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 252: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 252: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 253: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 253: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 254: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 254: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 255: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 255: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 256: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 256: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 257: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 257: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 258: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 258: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 259: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 259: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 260: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 260: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 261: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 261: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 262: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 262: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 263: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 263: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 264: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 264: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 265: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 265: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 266: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 266: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 267: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 267: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 268: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 268: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 269: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 269: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 270: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 270: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 271: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 271: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 272: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 272: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 273: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 273: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 274: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 274: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 275: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 275: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 276: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 276: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 277: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 277: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 278: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 278: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 279: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 279: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 280: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 280: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 281: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 281: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 282: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 282: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 283: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 283: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 284: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 284: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 285: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 285: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 286: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 286: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 287: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 287: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 288: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 288: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 289: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 289: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 290: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 290: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 291: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 291: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 292: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 292: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 293: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 293: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 294: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 294: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 295: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 295: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 296: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 296: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 297: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 297: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 298: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 298: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 299: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 299: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 100% 300/300 [53:20<00:00, 10.67s/it] INFO:root:Training is finished ... Iteration: 100% 24/24 [00:10<00:00, 2.32it/s, acc=1] Iteration: 100% 165/165 [23:52<00:00, 8.68s/it, acc=0.852] obj_pp_to_subj_pp: 15.1 cp_recursion: 51.5 pp_recursion: 46.9 subj_to_obj_proper: 94.8 prim_to_obj_proper: 70.9 prim_to_subj_proper: 100.0 LEX: 94.00666666666667 OVERALL: 85.20476190476191 INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 4344077 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.68 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:10<00:00, 19.94it/s, loss=5.94] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:10<00:00, 20.10it/s, loss=4.63] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:10<00:00, 20.09it/s, loss=3.63] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:10<00:00, 20.10it/s, loss=2.56] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:10<00:00, 20.08it/s, loss=2.02] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:10<00:00, 20.10it/s, loss=1.66] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:10<00:00, 20.02it/s, loss=1.43] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:10<00:00, 20.03it/s, loss=1.26] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:10<00:00, 20.07it/s, loss=1.15] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:10<00:00, 20.11it/s, loss=1.05] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.96] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0.88] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.81] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0.74] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0.67] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0.62] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0.53] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0.47] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0.41] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0.35] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0.3] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0.26] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0.23] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0.19] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0.18] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0.15] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0.14] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0.12] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0.12] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.09] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.09] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0.08] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0.07] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0.06] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0.06] Epoch: 35: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 35: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.05] Epoch: 36: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 36: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0.05] Epoch: 37: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 37: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0.05] Epoch: 38: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 38: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0.04] Epoch: 39: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 39: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0.03] Epoch: 40: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 40: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.04] Epoch: 41: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 41: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0.03] Epoch: 42: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 42: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0.03] Epoch: 43: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 43: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0.03] Epoch: 44: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 44: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0.03] Epoch: 45: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 45: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0.02] Epoch: 46: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 46: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0.03] Epoch: 47: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 47: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.04] Epoch: 48: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 48: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0.02] Epoch: 49: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 49: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0.02] Epoch: 50: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 50: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0.02] Epoch: 51: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 51: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0.02] Epoch: 52: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 52: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0.02] Epoch: 53: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 53: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0.03] Epoch: 54: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 54: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0.02] Epoch: 55: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 55: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0.02] Epoch: 56: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 56: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0.01] Epoch: 57: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 57: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0.02] Epoch: 58: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 58: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0.01] Epoch: 59: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 59: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0.01] Epoch: 60: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 60: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0.01] Epoch: 61: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 61: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0.01] Epoch: 62: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 62: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0.01] Epoch: 63: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 63: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0.01] Epoch: 64: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 64: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0.01] Epoch: 65: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 65: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0.01] Epoch: 66: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 66: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0.01] Epoch: 67: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 67: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0.01] Epoch: 68: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 68: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0.01] Epoch: 69: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 69: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.01] Epoch: 70: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 70: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 71: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 71: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0.01] Epoch: 72: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 72: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0.01] Epoch: 73: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 73: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0.01] Epoch: 74: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 74: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0.01] Epoch: 75: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 75: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0.01] Epoch: 76: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 76: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 77: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 77: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0.01] Epoch: 78: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 78: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0.01] Epoch: 79: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 79: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0.01] Epoch: 80: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 80: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0.01] Epoch: 81: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 81: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0.01] Epoch: 82: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 82: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 83: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 83: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 84: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 84: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 85: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 85: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0.01] Epoch: 86: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 86: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0.01] Epoch: 87: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 87: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 88: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 88: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0.01] Epoch: 89: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 89: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 90: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 90: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 91: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 91: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 92: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 92: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 93: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 93: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0.01] Epoch: 94: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 94: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 95: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 95: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0.01] Epoch: 96: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 96: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 97: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 97: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 98: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 98: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.01] Epoch: 99: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 99: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 100: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 100: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 101: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 101: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 102: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 102: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 103: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 103: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 104: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 104: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.01] Epoch: 105: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 105: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 106: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 106: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 107: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 107: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 108: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 108: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 109: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 109: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 110: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 110: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 111: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 111: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 112: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 112: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 113: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 113: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 114: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 114: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 115: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 115: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 116: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 116: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 117: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 117: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 118: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 118: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 119: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 119: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 120: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 120: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 121: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 121: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 122: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 122: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 123: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 123: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0.01] Epoch: 124: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 124: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 125: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 125: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 126: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 126: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 127: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 127: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 128: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 128: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 129: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 129: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 130: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 130: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 131: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 131: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 132: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 132: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 133: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 133: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 134: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 134: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 135: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 135: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 136: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 136: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 137: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 137: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 138: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 138: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 139: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 139: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 140: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 140: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 141: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 141: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 142: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 142: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 143: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 143: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 144: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 144: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 145: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 145: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 146: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 146: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 147: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 147: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 148: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 148: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 149: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 149: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 150: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 150: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 151: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 151: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 152: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 152: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 153: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 153: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 154: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 154: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 155: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 155: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 156: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 156: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 157: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 157: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 158: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 158: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 159: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 159: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 160: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 160: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 161: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 161: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 162: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 162: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 163: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 163: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 164: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 164: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 165: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 165: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 166: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 166: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 167: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 167: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 168: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 168: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 169: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 169: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 170: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 170: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 171: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 171: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 172: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 172: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 173: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 173: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 174: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 174: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 175: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 175: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 176: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 176: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 177: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 177: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 178: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 178: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 179: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 179: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 180: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 180: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 181: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 181: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 182: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 182: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 183: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 183: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 184: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 184: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 185: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 185: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 186: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 186: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 187: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 187: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 188: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 188: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 189: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 189: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 190: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 190: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 191: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 191: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 192: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 192: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 193: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 193: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 194: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 194: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 195: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 195: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 196: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 196: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 197: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 197: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 198: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 198: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 199: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 199: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 200: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 200: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 201: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 201: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 202: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 202: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 203: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 203: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 204: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 204: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 205: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 205: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 206: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 206: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 207: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 207: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 208: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 208: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 209: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 209: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 210: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 210: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 211: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 211: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 212: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 212: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 213: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 213: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 214: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 214: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 215: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 215: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 216: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 216: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 217: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 217: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 218: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 218: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 219: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 219: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 220: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 220: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 221: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 221: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 222: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 222: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 223: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 223: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 224: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 224: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 225: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 225: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 226: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 226: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 227: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 227: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 228: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 228: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 229: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 229: 100% 213/213 [00:10<00:00, 20.25it/s, loss=0] Epoch: 230: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 230: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 231: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 231: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 232: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 232: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 233: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 233: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 234: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 234: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 235: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 235: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 236: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 236: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 237: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 237: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 238: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 238: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 239: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 239: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 240: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 240: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 241: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 241: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 242: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 242: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 243: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 243: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 244: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 244: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 245: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 245: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 246: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 246: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 247: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 247: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 248: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 248: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 249: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 249: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 250: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 250: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 251: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 251: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 252: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 252: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 253: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 253: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 254: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 254: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 255: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 255: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 256: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 256: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 257: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 257: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 258: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 258: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 259: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 259: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 260: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 260: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 261: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 261: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 262: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 262: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 263: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 263: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 264: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 264: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 265: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 265: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 266: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 266: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 267: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 267: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 268: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 268: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 269: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 269: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 270: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 270: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 271: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 271: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 272: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 272: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 273: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 273: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 274: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 274: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 275: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 275: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 276: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 276: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 277: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 277: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 278: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 278: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 279: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 279: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 280: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 280: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 281: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 281: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 282: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 282: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 283: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 283: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 284: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 284: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 285: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 285: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 286: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 286: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 287: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 287: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 288: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 288: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 289: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 289: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 290: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 290: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 291: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 291: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 292: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 292: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 293: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 293: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 294: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 294: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 295: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 295: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 296: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 296: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 297: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 297: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 298: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 298: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 299: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 299: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 100% 300/300 [53:10<00:00, 10.64s/it] INFO:root:Training is finished ... Iteration: 100% 24/24 [00:10<00:00, 2.26it/s, acc=1] Iteration: 100% 165/165 [12:26<00:00, 4.53s/it, acc=0.903] obj_pp_to_subj_pp: 16.4 cp_recursion: 52.4 pp_recursion: 39.1 subj_to_obj_proper: 94.2 prim_to_obj_proper: 97.8 prim_to_subj_proper: 99.9 LEX: 99.80666666666667 OVERALL: 90.32857142857142 INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 4344077 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.79 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:10<00:00, 19.79it/s, loss=5.97] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:10<00:00, 19.87it/s, loss=4.63] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:10<00:00, 19.85it/s, loss=3.49] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:10<00:00, 19.80it/s, loss=2.41] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:10<00:00, 19.78it/s, loss=1.91] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:10<00:00, 19.84it/s, loss=1.58] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:10<00:00, 19.84it/s, loss=1.31] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:10<00:00, 19.84it/s, loss=1.11] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0.98] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0.87] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0.76] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0.68] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0.59] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0.51] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0.43] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0.37] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0.31] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0.25] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0.22] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0.18] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0.16] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0.14] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0.12] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0.12] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0.1] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0.08] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0.09] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0.07] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0.06] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0.06] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0.05] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0.04] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0.04] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0.04] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0.03] Epoch: 35: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 35: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0.03] Epoch: 36: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 36: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0.03] Epoch: 37: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 37: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0.03] Epoch: 38: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 38: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0.03] Epoch: 39: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 39: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0.03] Epoch: 40: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 40: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0.02] Epoch: 41: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 41: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0.02] Epoch: 42: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 42: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0.02] Epoch: 43: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 43: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0.02] Epoch: 44: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 44: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0.02] Epoch: 45: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 45: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0.02] Epoch: 46: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 46: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0.02] Epoch: 47: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 47: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0.02] Epoch: 48: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 48: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0.01] Epoch: 49: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 49: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0.01] Epoch: 50: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 50: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0.01] Epoch: 51: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 51: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0.01] Epoch: 52: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 52: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0.01] Epoch: 53: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 53: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0.01] Epoch: 54: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 54: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0.01] Epoch: 55: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 55: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0.01] Epoch: 56: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 56: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0.01] Epoch: 57: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 57: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0.01] Epoch: 58: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 58: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0.01] Epoch: 59: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 59: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0.01] Epoch: 60: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 60: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 61: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 61: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0.01] Epoch: 62: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 62: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0.01] Epoch: 63: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 63: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 64: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 64: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 65: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 65: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 66: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 66: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0.01] Epoch: 67: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 67: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 68: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 68: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0.01] Epoch: 69: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 69: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 70: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 70: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 71: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 71: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 72: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 72: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0] Epoch: 73: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 73: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0.01] Epoch: 74: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 74: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0.01] Epoch: 75: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 75: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0.01] Epoch: 76: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 76: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0.01] Epoch: 77: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 77: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 78: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 78: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0] Epoch: 79: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 79: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 80: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 80: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 81: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 81: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 82: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 82: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 83: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 83: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 84: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 84: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 85: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 85: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 86: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 86: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 87: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 87: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 88: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 88: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 89: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 89: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 90: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 90: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 91: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 91: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 92: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 92: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 93: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 93: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 94: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 94: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 95: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 95: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 96: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 96: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 97: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 97: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 98: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 98: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 99: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 99: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 100: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 100: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 101: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 101: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0] Epoch: 102: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 102: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 103: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 103: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 104: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 104: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 105: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 105: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 106: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 106: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 107: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 107: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 108: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 108: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 109: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 109: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 110: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 110: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 111: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 111: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 112: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 112: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 113: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 113: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 114: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 114: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 115: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 115: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 116: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 116: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 117: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 117: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 118: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 118: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 119: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 119: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 120: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 120: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 121: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 121: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 122: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 122: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 123: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 123: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 124: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 124: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 125: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 125: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 126: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 126: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 127: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 127: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 128: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 128: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 129: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 129: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 130: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 130: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0.01] Epoch: 131: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 131: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 132: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 132: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 133: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 133: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 134: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 134: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 135: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 135: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 136: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 136: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 137: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 137: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 138: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 138: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 139: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 139: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 140: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 140: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 141: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 141: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 142: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 142: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 143: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 143: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 144: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 144: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 145: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 145: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 146: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 146: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 147: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 147: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 148: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 148: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 149: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 149: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 150: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 150: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 151: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 151: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 152: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 152: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0] Epoch: 153: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 153: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 154: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 154: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 155: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 155: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 156: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 156: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 157: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 157: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 158: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 158: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 159: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 159: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 160: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 160: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 161: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 161: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 162: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 162: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 163: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 163: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0] Epoch: 164: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 164: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 165: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 165: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 166: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 166: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 167: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 167: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 168: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 168: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 169: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 169: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0] Epoch: 170: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 170: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 171: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 171: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 172: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 172: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 173: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 173: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 174: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 174: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 175: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 175: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 176: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 176: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 177: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 177: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 178: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 178: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 179: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 179: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 180: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 180: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 181: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 181: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 182: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 182: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 183: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 183: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0] Epoch: 184: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 184: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 185: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 185: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 186: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 186: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 187: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 187: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 188: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 188: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0] Epoch: 189: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 189: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 190: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 190: 100% 213/213 [00:10<00:00, 19.95it/s, loss=0] Epoch: 191: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 191: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 192: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 192: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 193: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 193: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 194: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 194: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 195: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 195: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 196: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 196: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 197: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 197: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 198: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 198: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 199: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 199: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 200: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 200: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 201: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 201: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 202: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 202: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 203: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 203: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 204: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 204: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 205: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 205: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 206: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 206: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0] Epoch: 207: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 207: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 208: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 208: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 209: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 209: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0] Epoch: 210: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 210: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0] Epoch: 211: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 211: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 212: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 212: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 213: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 213: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 214: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 214: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 215: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 215: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0] Epoch: 216: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 216: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 217: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 217: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 218: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 218: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 219: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 219: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 220: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 220: 100% 213/213 [00:10<00:00, 19.73it/s, loss=0] Epoch: 221: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 221: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 222: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 222: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 223: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 223: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 224: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 224: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 225: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 225: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 226: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 226: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 227: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 227: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 228: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 228: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 229: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 229: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 230: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 230: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 231: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 231: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 232: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 232: 100% 213/213 [00:10<00:00, 19.72it/s, loss=0] Epoch: 233: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 233: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0] Epoch: 234: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 234: 100% 213/213 [00:10<00:00, 19.69it/s, loss=0] Epoch: 235: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 235: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 236: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 236: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 237: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 237: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0] Epoch: 238: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 238: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 239: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 239: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 240: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 240: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 241: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 241: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 242: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 242: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0] Epoch: 243: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 243: 100% 213/213 [00:10<00:00, 19.72it/s, loss=0] Epoch: 244: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 244: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0] Epoch: 245: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 245: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 246: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 246: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 247: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 247: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 248: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 248: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 249: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 249: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0] Epoch: 250: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 250: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 251: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 251: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 252: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 252: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 253: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 253: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 254: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 254: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 255: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 255: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 256: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 256: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 257: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 257: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 258: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 258: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 259: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 259: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 260: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 260: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0] Epoch: 261: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 261: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 262: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 262: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0] Epoch: 263: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 263: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 264: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 264: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0] Epoch: 265: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 265: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 266: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 266: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0] Epoch: 267: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 267: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 268: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 268: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 269: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 269: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 270: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 270: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 271: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 271: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0] Epoch: 272: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 272: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 273: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 273: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 274: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 274: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 275: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 275: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 276: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 276: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 277: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 277: 100% 213/213 [00:10<00:00, 19.72it/s, loss=0] Epoch: 278: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 278: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 279: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 279: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 280: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 280: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 281: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 281: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 282: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 282: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 283: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 283: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0] Epoch: 284: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 284: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 285: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 285: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 286: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 286: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 287: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 287: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0] Epoch: 288: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 288: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0] Epoch: 289: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 289: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 290: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 290: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 291: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 291: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 292: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 292: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 293: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 293: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 294: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 294: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 295: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 295: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 296: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 296: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 297: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 297: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 298: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 298: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0] Epoch: 299: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 299: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 100% 300/300 [53:57<00:00, 10.79s/it] INFO:root:Training is finished ... Iteration: 100% 24/24 [00:10<00:00, 2.28it/s, acc=1] Iteration: 100% 165/165 [21:19<00:00, 7.76s/it, acc=0.909] obj_pp_to_subj_pp: 23.0 cp_recursion: 52.4 pp_recursion: 47.4 subj_to_obj_proper: 94.7 prim_to_obj_proper: 95.0 prim_to_subj_proper: 99.9 LEX: 99.72 OVERALL: 90.86666666666666 INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 4344077 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.90 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:10<00:00, 19.63it/s, loss=6.05] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:10<00:00, 19.90it/s, loss=4.62] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:10<00:00, 19.89it/s, loss=3.65] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:10<00:00, 19.88it/s, loss=2.55] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:10<00:00, 19.90it/s, loss=1.95] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:10<00:00, 19.83it/s, loss=1.62] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:10<00:00, 19.89it/s, loss=1.36] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:10<00:00, 19.94it/s, loss=1.16] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:10<00:00, 19.92it/s, loss=1.02] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0.92] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0.83] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0.76] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0.66] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0.59] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0.51] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0.44] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0.37] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0.31] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0.26] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0.24] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0.21] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0.18] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0.16] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0.14] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0.13] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0.12] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0.11] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0.09] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0.08] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0.08] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0.08] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0.08] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0.07] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0.06] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0.05] Epoch: 35: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 35: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0.05] Epoch: 36: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 36: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0.04] Epoch: 37: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 37: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0.03] Epoch: 38: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 38: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0.05] Epoch: 39: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 39: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0.03] Epoch: 40: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 40: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0.03] Epoch: 41: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 41: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0.03] Epoch: 42: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 42: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0.02] Epoch: 43: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 43: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0.03] Epoch: 44: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 44: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0.02] Epoch: 45: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 45: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0.02] Epoch: 46: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 46: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0.03] Epoch: 47: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 47: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0.02] Epoch: 48: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 48: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0.02] Epoch: 49: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 49: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0.02] Epoch: 50: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 50: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0.02] Epoch: 51: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 51: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0.01] Epoch: 52: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 52: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0.02] Epoch: 53: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 53: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0.01] Epoch: 54: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 54: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0.01] Epoch: 55: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 55: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0.01] Epoch: 56: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 56: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0.01] Epoch: 57: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 57: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0.01] Epoch: 58: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 58: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0.02] Epoch: 59: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 59: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0.01] Epoch: 60: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 60: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0.01] Epoch: 61: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 61: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0.01] Epoch: 62: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 62: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0.01] Epoch: 63: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 63: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0.01] Epoch: 64: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 64: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0.01] Epoch: 65: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 65: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0.01] Epoch: 66: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 66: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0.01] Epoch: 67: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 67: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0.01] Epoch: 68: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 68: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0.01] Epoch: 69: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 69: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0.01] Epoch: 70: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 70: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0.01] Epoch: 71: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 71: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0.01] Epoch: 72: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 72: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0.01] Epoch: 73: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 73: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0.01] Epoch: 74: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 74: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0.01] Epoch: 75: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 75: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0.01] Epoch: 76: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 76: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0.01] Epoch: 77: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 77: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0.01] Epoch: 78: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 78: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0.01] Epoch: 79: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 79: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0.01] Epoch: 80: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 80: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0.01] Epoch: 81: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 81: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0.01] Epoch: 82: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 82: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0.01] Epoch: 83: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 83: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0.01] Epoch: 84: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 84: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0.01] Epoch: 85: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 85: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0] Epoch: 86: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 86: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0.01] Epoch: 87: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 87: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0.01] Epoch: 88: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 88: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0.01] Epoch: 89: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 89: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0.01] Epoch: 90: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 90: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0.01] Epoch: 91: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 91: 100% 213/213 [00:10<00:00, 19.73it/s, loss=0] Epoch: 92: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 92: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 93: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 93: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 94: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 94: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 95: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 95: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 96: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 96: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0.01] Epoch: 97: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 97: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 98: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 98: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 99: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 99: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0.01] Epoch: 100: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 100: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0.01] Epoch: 101: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 101: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 102: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 102: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 103: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 103: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 104: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 104: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 105: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 105: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 106: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 106: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 107: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 107: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 108: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 108: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 109: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 109: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 110: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 110: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0.01] Epoch: 111: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 111: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 112: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 112: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 113: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 113: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0.01] Epoch: 114: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 114: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 115: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 115: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 116: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 116: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 117: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 117: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 118: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 118: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 119: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 119: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0] Epoch: 120: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 120: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 121: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 121: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 122: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 122: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 123: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 123: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 124: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 124: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 125: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 125: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 126: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 126: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 127: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 127: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 128: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 128: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 129: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 129: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 130: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 130: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 131: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 131: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 132: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 132: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 133: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 133: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 134: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 134: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0] Epoch: 135: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 135: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 136: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 136: 100% 213/213 [00:10<00:00, 19.69it/s, loss=0] Epoch: 137: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 137: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0] Epoch: 138: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 138: 100% 213/213 [00:10<00:00, 19.66it/s, loss=0] Epoch: 139: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 139: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0] Epoch: 140: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 140: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 141: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 141: 100% 213/213 [00:10<00:00, 19.73it/s, loss=0] Epoch: 142: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 142: 100% 213/213 [00:10<00:00, 19.72it/s, loss=0] Epoch: 143: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 143: 100% 213/213 [00:10<00:00, 19.71it/s, loss=0] Epoch: 144: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 144: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0] Epoch: 145: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 145: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 146: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 146: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 147: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 147: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 148: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 148: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 149: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 149: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 150: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 150: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 151: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 151: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 152: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 152: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0] Epoch: 153: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 153: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0] Epoch: 154: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 154: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 155: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 155: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 156: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 156: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0] Epoch: 157: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 157: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 158: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 158: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 159: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 159: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0] Epoch: 160: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 160: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 161: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 161: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 162: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 162: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 163: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 163: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 164: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 164: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 165: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 165: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 166: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 166: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 167: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 167: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 168: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 168: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 169: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 169: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 170: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 170: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 171: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 171: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 172: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 172: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 173: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 173: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 174: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 174: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 175: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 175: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 176: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 176: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 177: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 177: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 178: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 178: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 179: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 179: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 180: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 180: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 181: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 181: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 182: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 182: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0] Epoch: 183: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 183: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 184: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 184: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 185: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 185: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 186: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 186: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 187: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 187: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 188: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 188: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 189: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 189: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 190: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 190: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 191: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 191: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 192: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 192: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 193: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 193: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0] Epoch: 194: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 194: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 195: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 195: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0] Epoch: 196: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 196: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 197: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 197: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 198: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 198: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 199: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 199: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 200: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 200: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 201: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 201: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 202: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 202: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 203: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 203: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0] Epoch: 204: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 204: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 205: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 205: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0] Epoch: 206: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 206: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 207: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 207: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 208: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 208: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0] Epoch: 209: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 209: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 210: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 210: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0] Epoch: 211: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 211: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 212: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 212: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 213: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 213: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 214: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 214: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 215: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 215: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 216: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 216: 100% 213/213 [00:10<00:00, 19.73it/s, loss=0] Epoch: 217: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 217: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 218: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 218: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 219: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 219: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 220: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 220: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 221: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 221: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 222: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 222: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 223: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 223: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 224: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 224: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0] Epoch: 225: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 225: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0] Epoch: 226: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 226: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 227: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 227: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0] Epoch: 228: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 228: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 229: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 229: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 230: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 230: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 231: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 231: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 232: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 232: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 233: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 233: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 234: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 234: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 235: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 235: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 236: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 236: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 237: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 237: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 238: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 238: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 239: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 239: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 240: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 240: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 241: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 241: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 242: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 242: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 243: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 243: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 244: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 244: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 245: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 245: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 246: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 246: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 247: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 247: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 248: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 248: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 249: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 249: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 250: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 250: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 251: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 251: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 252: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 252: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 253: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 253: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 254: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 254: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 255: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 255: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 256: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 256: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 257: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 257: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 258: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 258: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 259: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 259: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 260: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 260: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 261: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 261: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0] Epoch: 262: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 262: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 263: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 263: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 264: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 264: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 265: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 265: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 266: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 266: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 267: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 267: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0] Epoch: 268: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 268: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 269: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 269: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 270: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 270: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 271: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 271: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 272: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 272: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 273: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 273: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0] Epoch: 274: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 274: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 275: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 275: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 276: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 276: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 277: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 277: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 278: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 278: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 279: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 279: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 280: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 280: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 281: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 281: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0] Epoch: 282: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 282: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 283: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 283: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0] Epoch: 284: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 284: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 285: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 285: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 286: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 286: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 287: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 287: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 288: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 288: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 289: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 289: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 290: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 290: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 291: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 291: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 292: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 292: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 293: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 293: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 294: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 294: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 295: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 295: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 296: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 296: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 297: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 297: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 298: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 298: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 299: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 299: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 100% 300/300 [53:58<00:00, 10.80s/it] INFO:root:Training is finished ... Iteration: 100% 24/24 [00:10<00:00, 2.26it/s, acc=1] Iteration: 100% 165/165 [18:55<00:00, 6.88s/it, acc=0.901] obj_pp_to_subj_pp: 12.7 cp_recursion: 52.0 pp_recursion: 37.3 subj_to_obj_proper: 95.5 prim_to_obj_proper: 98.2 prim_to_subj_proper: 99.9 LEX: 99.76666666666667 OVERALL: 90.10000000000001 INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 4344077 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.101 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:10<00:00, 19.77it/s, loss=5.98] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:10<00:00, 20.00it/s, loss=4.56] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:10<00:00, 20.01it/s, loss=3.49] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:10<00:00, 19.95it/s, loss=2.42] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:10<00:00, 19.99it/s, loss=1.93] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:10<00:00, 20.02it/s, loss=1.57] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:10<00:00, 20.00it/s, loss=1.32] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:10<00:00, 20.04it/s, loss=1.15] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:10<00:00, 20.01it/s, loss=1.02] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0.92] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0.84] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0.76] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0.67] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0.58] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0.51] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0.43] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0.37] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0.32] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0.26] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0.22] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0.19] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0.15] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0.14] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0.12] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0.09] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0.09] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0.07] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0.06] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0.05] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0.05] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0.04] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0.04] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0.04] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0.04] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0.02] Epoch: 35: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 35: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0.02] Epoch: 36: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 36: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0.02] Epoch: 37: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 37: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0.02] Epoch: 38: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 38: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0.02] Epoch: 39: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 39: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0.02] Epoch: 40: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 40: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0.01] Epoch: 41: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 41: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0.01] Epoch: 42: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 42: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0.01] Epoch: 43: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 43: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0.01] Epoch: 44: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 44: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0.01] Epoch: 45: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 45: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0.01] Epoch: 46: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 46: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0.01] Epoch: 47: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 47: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0.01] Epoch: 48: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 48: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0.02] Epoch: 49: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 49: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0.01] Epoch: 50: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 50: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0.01] Epoch: 51: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 51: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0.01] Epoch: 52: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 52: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0.01] Epoch: 53: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 53: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0.01] Epoch: 54: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 54: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0.01] Epoch: 55: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 55: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0.01] Epoch: 56: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 56: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0.01] Epoch: 57: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 57: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 58: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 58: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0.01] Epoch: 59: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 59: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 60: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 60: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 61: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 61: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 62: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 62: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 63: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 63: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 64: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 64: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 65: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 65: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 66: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 66: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 67: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 67: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 68: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 68: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 69: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 69: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 70: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 70: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 71: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 71: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 72: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 72: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 73: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 73: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 74: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 74: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 75: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 75: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 76: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 76: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 77: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 77: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 78: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 78: 100% 213/213 [00:10<00:00, 19.95it/s, loss=0] Epoch: 79: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 79: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 80: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 80: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 81: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 81: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 82: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 82: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 83: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 83: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0] Epoch: 84: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 84: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 85: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 85: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 86: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 86: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 87: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 87: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 88: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 88: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 89: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 89: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0] Epoch: 90: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 90: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 91: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 91: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 92: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 92: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 93: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 93: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0] Epoch: 94: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 94: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0] Epoch: 95: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 95: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 96: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 96: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 97: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 97: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 98: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 98: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 99: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 99: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0] Epoch: 100: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 100: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 101: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 101: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 102: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 102: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 103: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 103: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 104: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 104: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 105: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 105: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0.01] Epoch: 106: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 106: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 107: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 107: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 108: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 108: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 109: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 109: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0] Epoch: 110: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 110: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 111: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 111: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 112: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 112: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 113: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 113: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 114: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 114: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 115: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 115: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0] Epoch: 116: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 116: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 117: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 117: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 118: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 118: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 119: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 119: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 120: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 120: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 121: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 121: 100% 213/213 [00:10<00:00, 19.95it/s, loss=0] Epoch: 122: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 122: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 123: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 123: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 124: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 124: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0] Epoch: 125: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 125: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 126: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 126: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 127: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 127: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 128: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 128: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 129: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 129: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 130: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 130: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 131: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 131: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0] Epoch: 132: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 132: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0] Epoch: 133: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 133: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0] Epoch: 134: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 134: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 135: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 135: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 136: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 136: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 137: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 137: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 138: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 138: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0] Epoch: 139: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 139: 100% 213/213 [00:10<00:00, 19.95it/s, loss=0] Epoch: 140: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 140: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 141: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 141: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 142: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 142: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0] Epoch: 143: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 143: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0] Epoch: 144: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 144: 100% 213/213 [00:10<00:00, 19.95it/s, loss=0] Epoch: 145: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 145: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 146: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 146: 100% 213/213 [00:10<00:00, 19.73it/s, loss=0] Epoch: 147: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 147: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 148: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 148: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 149: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 149: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 150: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 150: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 151: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 151: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 152: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 152: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 153: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 153: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0] Epoch: 154: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 154: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 155: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 155: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 156: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 156: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 157: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 157: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 158: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 158: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 159: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 159: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 160: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 160: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 161: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 161: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 162: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 162: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 163: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 163: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 164: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 164: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 165: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 165: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 166: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 166: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 167: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 167: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0] Epoch: 168: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 168: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 169: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 169: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 170: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 170: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 171: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 171: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 172: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 172: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 173: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 173: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 174: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 174: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 175: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 175: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 176: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 176: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 177: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 177: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 178: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 178: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 179: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 179: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 180: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 180: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 181: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 181: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0] Epoch: 182: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 182: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 183: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 183: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 184: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 184: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 185: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 185: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 186: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 186: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 187: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 187: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 188: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 188: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 189: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 189: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 190: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 190: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 191: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 191: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0] Epoch: 192: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 192: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 193: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 193: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0] Epoch: 194: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 194: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 195: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 195: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0] Epoch: 196: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 196: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0] Epoch: 197: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 197: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 198: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 198: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 199: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 199: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 200: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 200: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 201: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 201: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0] Epoch: 202: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 202: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 203: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 203: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 204: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 204: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0] Epoch: 205: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 205: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 206: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 206: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 207: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 207: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 208: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 208: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 209: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 209: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 210: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 210: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 211: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 211: 100% 213/213 [00:10<00:00, 19.95it/s, loss=0] Epoch: 212: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 212: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 213: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 213: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 214: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 214: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 215: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 215: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 216: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 216: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 217: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 217: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 218: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 218: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 219: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 219: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 220: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 220: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 221: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 221: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 222: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 222: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 223: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 223: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 224: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 224: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 225: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 225: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 226: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 226: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 227: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 227: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 228: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 228: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0] Epoch: 229: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 229: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 230: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 230: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 231: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 231: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 232: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 232: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 233: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 233: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 234: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 234: 100% 213/213 [00:10<00:00, 19.95it/s, loss=0] Epoch: 235: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 235: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 236: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 236: 100% 213/213 [00:10<00:00, 19.95it/s, loss=0] Epoch: 237: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 237: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 238: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 238: 100% 213/213 [00:10<00:00, 19.95it/s, loss=0] Epoch: 239: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 239: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 240: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 240: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 241: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 241: 100% 213/213 [00:10<00:00, 19.95it/s, loss=0] Epoch: 242: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 242: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0] Epoch: 243: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 243: 100% 213/213 [00:10<00:00, 19.95it/s, loss=0] Epoch: 244: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 244: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 245: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 245: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 246: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 246: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0] Epoch: 247: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 247: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 248: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 248: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 249: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 249: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 250: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 250: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 251: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 251: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 252: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 252: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 253: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 253: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 254: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 254: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 255: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 255: 100% 213/213 [00:10<00:00, 19.95it/s, loss=0] Epoch: 256: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 256: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 257: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 257: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 258: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 258: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 259: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 259: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 260: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 260: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 261: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 261: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0] Epoch: 262: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 262: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 263: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 263: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 264: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 264: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0] Epoch: 265: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 265: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0] Epoch: 266: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 266: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 267: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 267: 100% 213/213 [00:10<00:00, 19.95it/s, loss=0] Epoch: 268: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 268: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 269: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 269: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 270: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 270: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 271: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 271: 100% 213/213 [00:10<00:00, 19.95it/s, loss=0] Epoch: 272: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 272: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 273: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 273: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 274: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 274: 100% 213/213 [00:10<00:00, 19.95it/s, loss=0] Epoch: 275: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 275: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 276: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 276: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0] Epoch: 277: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 277: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0] Epoch: 278: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 278: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 279: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 279: 100% 213/213 [00:10<00:00, 19.95it/s, loss=0] Epoch: 280: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 280: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0] Epoch: 281: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 281: 100% 213/213 [00:10<00:00, 19.95it/s, loss=0] Epoch: 282: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 282: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 283: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 283: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 284: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 284: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 285: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 285: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0] Epoch: 286: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 286: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 287: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 287: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0] Epoch: 288: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 288: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0] Epoch: 289: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 289: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 290: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 290: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 291: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 291: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 292: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 292: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 293: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 293: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 294: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 294: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 295: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 295: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 296: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 296: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 297: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 297: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 298: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 298: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0] Epoch: 299: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 299: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0] Epoch: 100% 300/300 [53:40<00:00, 10.74s/it] INFO:root:Training is finished ... Iteration: 100% 24/24 [00:10<00:00, 2.26it/s, acc=1] Iteration: 100% 165/165 [19:30<00:00, 7.09s/it, acc=0.894] obj_pp_to_subj_pp: 12.4 cp_recursion: 51.1 pp_recursion: 46.7 subj_to_obj_proper: 90.0 prim_to_obj_proper: 90.1 prim_to_subj_proper: 100.0 LEX: 99.14 OVERALL: 89.4
In [ ]:
!python run_cogs.py --model_name ende_transformer --gpu 1 --train_batch_size 128 --eval_batch_size 128 --lr 0.0001 --data_path ./recogs_positional_index --output_dir ./results_recogs_positional_index --lfs cogs --do_train --do_test --do_gen --max_seq_len 512 --output_json --epochs 300 --seeds "45;69;80;91;102" # paper's seeds + 3, not sure how they chose them
INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 4344077 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.45 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]We strongly recommend passing in an `attention_mask` since your input_ids may be padded. See https://huggingface.co/docs/transformers/troubleshooting#incorrect-output-when-padding-tokens-arent-masked. /content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:11<00:00, 18.49it/s, loss=5.99] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:10<00:00, 20.01it/s, loss=4.65] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:10<00:00, 19.98it/s, loss=3.61] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:10<00:00, 19.93it/s, loss=2.52] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:10<00:00, 20.02it/s, loss=1.95] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:10<00:00, 19.98it/s, loss=1.59] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:10<00:00, 19.96it/s, loss=1.32] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:10<00:00, 19.93it/s, loss=1.13] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:10<00:00, 19.89it/s, loss=1.01] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0.91] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:10<00:00, 19.95it/s, loss=0.8] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0.72] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0.62] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0.53] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0.46] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0.39] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0.33] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0.28] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0.24] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0.21] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0.18] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0.15] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0.13] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0.12] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0.12] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0.09] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0.08] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0.07] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0.06] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0.08] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0.05] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0.05] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0.04] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0.03] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0.04] Epoch: 35: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 35: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0.03] Epoch: 36: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 36: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0.03] Epoch: 37: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 37: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0.03] Epoch: 38: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 38: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0.03] Epoch: 39: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 39: 100% 213/213 [00:10<00:00, 19.95it/s, loss=0.02] Epoch: 40: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 40: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0.02] Epoch: 41: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 41: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0.02] Epoch: 42: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 42: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0.03] Epoch: 43: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 43: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0.01] Epoch: 44: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 44: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0.02] Epoch: 45: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 45: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0.02] Epoch: 46: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 46: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0.02] Epoch: 47: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 47: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0.01] Epoch: 48: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 48: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0.01] Epoch: 49: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 49: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0.01] Epoch: 50: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 50: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0.01] Epoch: 51: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 51: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0.01] Epoch: 52: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 52: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0.01] Epoch: 53: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 53: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0.01] Epoch: 54: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 54: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0.01] Epoch: 55: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 55: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0.01] Epoch: 56: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 56: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0.01] Epoch: 57: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 57: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0.01] Epoch: 58: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 58: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0.01] Epoch: 59: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 59: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0.01] Epoch: 60: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 60: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0.01] Epoch: 61: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 61: 100% 213/213 [00:10<00:00, 19.95it/s, loss=0] Epoch: 62: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 62: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0.01] Epoch: 63: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 63: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 64: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 64: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0.01] Epoch: 65: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 65: 100% 213/213 [00:10<00:00, 19.95it/s, loss=0.01] Epoch: 66: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 66: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 67: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 67: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0] Epoch: 68: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 68: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 69: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 69: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 70: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 70: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 71: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 71: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0.01] Epoch: 72: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 72: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 73: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 73: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 74: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 74: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 75: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 75: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 76: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 76: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 77: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 77: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 78: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 78: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 79: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 79: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 80: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 80: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0.01] Epoch: 81: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 81: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 82: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 82: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 83: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 83: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 84: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 84: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 85: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 85: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 86: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 86: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 87: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 87: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 88: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 88: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 89: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 89: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 90: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 90: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 91: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 91: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 92: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 92: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 93: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 93: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 94: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 94: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0] Epoch: 95: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 95: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 96: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 96: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0] Epoch: 97: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 97: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 98: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 98: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 99: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 99: 100% 213/213 [00:10<00:00, 19.95it/s, loss=0] Epoch: 100: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 100: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 101: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 101: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 102: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 102: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 103: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 103: 100% 213/213 [00:10<00:00, 19.95it/s, loss=0] Epoch: 104: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 104: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 105: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 105: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0] Epoch: 106: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 106: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 107: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 107: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 108: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 108: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 109: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 109: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 110: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 110: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 111: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 111: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 112: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 112: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 113: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 113: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 114: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 114: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 115: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 115: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 116: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 116: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 117: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 117: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 118: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 118: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 119: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 119: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 120: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 120: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0] Epoch: 121: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 121: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 122: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 122: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 123: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 123: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 124: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 124: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 125: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 125: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 126: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 126: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 127: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 127: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 128: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 128: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 129: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 129: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 130: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 130: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 131: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 131: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 132: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 132: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0] Epoch: 133: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 133: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 134: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 134: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 135: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 135: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 136: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 136: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 137: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 137: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 138: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 138: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 139: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 139: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 140: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 140: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 141: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 141: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 142: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 142: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 143: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 143: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 144: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 144: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 145: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 145: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 146: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 146: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 147: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 147: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 148: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 148: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 149: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 149: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 150: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 150: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 151: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 151: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 152: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 152: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 153: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 153: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 154: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 154: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 155: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 155: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 156: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 156: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 157: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 157: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 158: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 158: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 159: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 159: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 160: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 160: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 161: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 161: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 162: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 162: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 163: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 163: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 164: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 164: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 165: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 165: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 166: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 166: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 167: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 167: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 168: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 168: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 169: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 169: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 170: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 170: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 171: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 171: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 172: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 172: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 173: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 173: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 174: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 174: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 175: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 175: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 176: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 176: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 177: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 177: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 178: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 178: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 179: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 179: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 180: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 180: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 181: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 181: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 182: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 182: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 183: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 183: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 184: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 184: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 185: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 185: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 186: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 186: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 187: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 187: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 188: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 188: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 189: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 189: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 190: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 190: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 191: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 191: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 192: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 192: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0] Epoch: 193: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 193: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 194: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 194: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 195: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 195: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 196: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 196: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 197: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 197: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 198: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 198: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 199: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 199: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 200: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 200: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 201: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 201: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 202: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 202: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 203: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 203: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 204: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 204: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 205: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 205: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 206: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 206: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 207: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 207: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 208: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 208: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 209: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 209: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 210: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 210: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 211: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 211: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 212: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 212: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 213: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 213: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 214: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 214: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 215: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 215: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 216: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 216: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 217: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 217: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 218: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 218: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 219: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 219: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 220: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 220: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 221: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 221: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 222: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 222: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 223: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 223: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 224: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 224: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 225: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 225: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 226: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 226: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 227: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 227: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 228: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 228: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 229: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 229: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 230: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 230: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 231: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 231: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 232: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 232: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 233: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 233: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 234: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 234: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 235: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 235: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 236: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 236: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 237: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 237: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 238: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 238: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0] Epoch: 239: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 239: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 240: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 240: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 241: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 241: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 242: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 242: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 243: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 243: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 244: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 244: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 245: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 245: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 246: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 246: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 247: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 247: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 248: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 248: 100% 213/213 [00:10<00:00, 19.95it/s, loss=0] Epoch: 249: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 249: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 250: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 250: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 251: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 251: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 252: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 252: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 253: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 253: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 254: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 254: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 255: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 255: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 256: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 256: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 257: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 257: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 258: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 258: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 259: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 259: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 260: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 260: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 261: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 261: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0] Epoch: 262: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 262: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 263: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 263: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 264: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 264: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 265: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 265: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 266: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 266: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 267: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 267: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0] Epoch: 268: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 268: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 269: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 269: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 270: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 270: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 271: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 271: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 272: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 272: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 273: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 273: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 274: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 274: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 275: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 275: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 276: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 276: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 277: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 277: 100% 213/213 [00:10<00:00, 19.95it/s, loss=0] Epoch: 278: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 278: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 279: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 279: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 280: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 280: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 281: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 281: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 282: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 282: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 283: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 283: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 284: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 284: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 285: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 285: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 286: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 286: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 287: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 287: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 288: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 288: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 289: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 289: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 290: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 290: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 291: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 291: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 292: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 292: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0] Epoch: 293: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 293: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 294: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 294: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 295: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 295: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 296: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 296: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 297: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 297: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 298: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 298: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 299: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 299: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 100% 300/300 [53:36<00:00, 10.72s/it] INFO:root:Training is finished ... Iteration: 100% 24/24 [00:10<00:00, 2.35it/s, acc=1] Iteration: 100% 165/165 [23:20<00:00, 8.49s/it, acc=0.907] obj_pp_to_subj_pp: 31.4 cp_recursion: 55.9 pp_recursion: 34.4 subj_to_obj_proper: 95.7 prim_to_obj_proper: 88.0 prim_to_subj_proper: 99.8 LEX: 99.93333333333332 OVERALL: 90.67619047619047 INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 4344077 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.69 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:10<00:00, 19.66it/s, loss=5.98] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:10<00:00, 19.90it/s, loss=4.61] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:10<00:00, 19.92it/s, loss=3.6] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:10<00:00, 19.93it/s, loss=2.51] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:10<00:00, 19.81it/s, loss=1.91] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:10<00:00, 19.86it/s, loss=1.55] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:10<00:00, 19.91it/s, loss=1.28] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:10<00:00, 19.87it/s, loss=1.09] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0.95] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0.84] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0.74] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0.66] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0.57] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0.48] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0.4] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0.33] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0.28] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0.23] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0.19] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0.16] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0.14] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0.11] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0.09] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0.07] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0.07] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0.06] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0.05] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0.04] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0.04] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0.04] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0.03] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0.03] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0.03] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0.02] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0.02] Epoch: 35: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 35: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0.02] Epoch: 36: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 36: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0.02] Epoch: 37: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 37: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0.01] Epoch: 38: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 38: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0.01] Epoch: 39: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 39: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0.02] Epoch: 40: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 40: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0.01] Epoch: 41: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 41: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0.01] Epoch: 42: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 42: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0.02] Epoch: 43: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 43: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0.01] Epoch: 44: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 44: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0.01] Epoch: 45: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 45: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0.01] Epoch: 46: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 46: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0.01] Epoch: 47: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 47: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0] Epoch: 48: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 48: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0.01] Epoch: 49: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 49: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0.01] Epoch: 50: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 50: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0.01] Epoch: 51: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 51: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0.01] Epoch: 52: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 52: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 53: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 53: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 54: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 54: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0] Epoch: 55: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 55: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 56: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 56: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 57: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 57: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0] Epoch: 58: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 58: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 59: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 59: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 60: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 60: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 61: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 61: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 62: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 62: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 63: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 63: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0] Epoch: 64: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 64: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 65: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 65: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0] Epoch: 66: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 66: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 67: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 67: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 68: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 68: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 69: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 69: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0] Epoch: 70: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 70: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 71: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 71: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 72: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 72: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 73: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 73: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 74: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 74: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 75: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 75: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 76: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 76: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 77: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 77: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0] Epoch: 78: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 78: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 79: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 79: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 80: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 80: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 81: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 81: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 82: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 82: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 83: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 83: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 84: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 84: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 85: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 85: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 86: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 86: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 87: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 87: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 88: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 88: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 89: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 89: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 90: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 90: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 91: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 91: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 92: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 92: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 93: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 93: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 94: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 94: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 95: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 95: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 96: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 96: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0] Epoch: 97: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 97: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 98: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 98: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 99: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 99: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 100: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 100: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 101: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 101: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0] Epoch: 102: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 102: 100% 213/213 [00:10<00:00, 19.95it/s, loss=0] Epoch: 103: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 103: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 104: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 104: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 105: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 105: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 106: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 106: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 107: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 107: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 108: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 108: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 109: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 109: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 110: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 110: 100% 213/213 [00:10<00:00, 19.71it/s, loss=0] Epoch: 111: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 111: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0] Epoch: 112: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 112: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 113: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 113: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 114: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 114: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 115: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 115: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 116: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 116: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 117: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 117: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0] Epoch: 118: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 118: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 119: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 119: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 120: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 120: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 121: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 121: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 122: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 122: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0] Epoch: 123: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 123: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 124: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 124: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 125: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 125: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 126: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 126: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 127: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 127: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 128: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 128: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 129: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 129: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 130: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 130: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 131: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 131: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 132: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 132: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 133: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 133: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0] Epoch: 134: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 134: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 135: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 135: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 136: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 136: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 137: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 137: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 138: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 138: 100% 213/213 [00:10<00:00, 19.73it/s, loss=0] Epoch: 139: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 139: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 140: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 140: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0] Epoch: 141: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 141: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 142: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 142: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 143: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 143: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 144: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 144: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 145: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 145: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 146: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 146: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 147: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 147: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 148: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 148: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 149: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 149: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0] Epoch: 150: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 150: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0] Epoch: 151: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 151: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 152: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 152: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 153: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 153: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 154: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 154: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 155: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 155: 100% 213/213 [00:10<00:00, 19.73it/s, loss=0] Epoch: 156: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 156: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0] Epoch: 157: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 157: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0] Epoch: 158: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 158: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0] Epoch: 159: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 159: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 160: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 160: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0] Epoch: 161: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 161: 100% 213/213 [00:10<00:00, 19.69it/s, loss=0] Epoch: 162: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 162: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 163: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 163: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 164: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 164: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 165: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 165: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 166: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 166: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 167: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 167: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 168: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 168: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 169: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 169: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 170: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 170: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 171: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 171: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 172: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 172: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 173: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 173: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 174: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 174: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 175: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 175: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 176: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 176: 100% 213/213 [00:10<00:00, 19.73it/s, loss=0] Epoch: 177: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 177: 100% 213/213 [00:10<00:00, 19.72it/s, loss=0] Epoch: 178: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 178: 100% 213/213 [00:10<00:00, 19.68it/s, loss=0] Epoch: 179: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 179: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0] Epoch: 180: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 180: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 181: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 181: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 182: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 182: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0] Epoch: 183: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 183: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 184: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 184: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0] Epoch: 185: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 185: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 186: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 186: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 187: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 187: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 188: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 188: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 189: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 189: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0] Epoch: 190: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 190: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0] Epoch: 191: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 191: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 192: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 192: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 193: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 193: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 194: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 194: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 195: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 195: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 196: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 196: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 197: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 197: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 198: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 198: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 199: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 199: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 200: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 200: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 201: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 201: 100% 213/213 [00:10<00:00, 19.72it/s, loss=0] Epoch: 202: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 202: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0] Epoch: 203: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 203: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 204: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 204: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 205: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 205: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 206: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 206: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0] Epoch: 207: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 207: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0] Epoch: 208: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 208: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0] Epoch: 209: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 209: 100% 213/213 [00:10<00:00, 19.73it/s, loss=0] Epoch: 210: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 210: 100% 213/213 [00:10<00:00, 19.73it/s, loss=0] Epoch: 211: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 211: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 212: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 212: 100% 213/213 [00:10<00:00, 19.73it/s, loss=0] Epoch: 213: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 213: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 214: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 214: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 215: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 215: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 216: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 216: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 217: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 217: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 218: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 218: 100% 213/213 [00:10<00:00, 19.72it/s, loss=0] Epoch: 219: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 219: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 220: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 220: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 221: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 221: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 222: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 222: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 223: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 223: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 224: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 224: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 225: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 225: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 226: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 226: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 227: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 227: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 228: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 228: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 229: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 229: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0] Epoch: 230: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 230: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 231: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 231: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 232: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 232: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 233: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 233: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 234: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 234: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 235: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 235: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0] Epoch: 236: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 236: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 237: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 237: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 238: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 238: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 239: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 239: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 240: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 240: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 241: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 241: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 242: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 242: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 243: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 243: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 244: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 244: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 245: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 245: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 246: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 246: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 247: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 247: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 248: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 248: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 249: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 249: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 250: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 250: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 251: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 251: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 252: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 252: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 253: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 253: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 254: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 254: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 255: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 255: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 256: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 256: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 257: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 257: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 258: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 258: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0] Epoch: 259: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 259: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 260: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 260: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 261: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 261: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 262: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 262: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 263: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 263: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0] Epoch: 264: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 264: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 265: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 265: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 266: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 266: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 267: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 267: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 268: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 268: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 269: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 269: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 270: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 270: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 271: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 271: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 272: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 272: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 273: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 273: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 274: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 274: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 275: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 275: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 276: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 276: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 277: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 277: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 278: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 278: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 279: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 279: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 280: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 280: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 281: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 281: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 282: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 282: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 283: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 283: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 284: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 284: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 285: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 285: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 286: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 286: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 287: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 287: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 288: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 288: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 289: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 289: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 290: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 290: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 291: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 291: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 292: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 292: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 293: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 293: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 294: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 294: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 295: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 295: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 296: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 296: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 297: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 297: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 298: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 298: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 299: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 299: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 100% 300/300 [53:56<00:00, 10.79s/it] INFO:root:Training is finished ... Iteration: 100% 24/24 [00:10<00:00, 2.27it/s, acc=1] Iteration: 100% 165/165 [23:33<00:00, 8.57s/it, acc=0.892] obj_pp_to_subj_pp: 35.3 cp_recursion: 52.9 pp_recursion: 36.9 subj_to_obj_proper: 87.5 prim_to_obj_proper: 66.5 prim_to_subj_proper: 99.9 LEX: 99.62 OVERALL: 89.20476190476191 INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 4344077 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.80 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:10<00:00, 19.93it/s, loss=5.97] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:10<00:00, 20.14it/s, loss=4.58] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:10<00:00, 20.05it/s, loss=3.5] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:10<00:00, 20.10it/s, loss=2.44] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:10<00:00, 20.10it/s, loss=1.88] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:10<00:00, 20.12it/s, loss=1.53] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:10<00:00, 19.96it/s, loss=1.29] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:10<00:00, 20.11it/s, loss=1.1] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0.97] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0.88] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0.79] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0.71] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0.64] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0.57] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.5] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.43] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0.39] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0.34] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0.3] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0.26] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.22] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0.2] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0.17] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0.15] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0.13] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.12] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0.12] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0.1] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0.09] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0.09] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0.08] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0.06] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0.07] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0.06] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.06] Epoch: 35: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 35: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0.05] Epoch: 36: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 36: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.06] Epoch: 37: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 37: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0.04] Epoch: 38: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 38: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0.04] Epoch: 39: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 39: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.04] Epoch: 40: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 40: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.03] Epoch: 41: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 41: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0.03] Epoch: 42: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 42: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.04] Epoch: 43: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 43: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0.03] Epoch: 44: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 44: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0.03] Epoch: 45: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 45: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.03] Epoch: 46: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 46: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0.03] Epoch: 47: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 47: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0.03] Epoch: 48: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 48: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0.02] Epoch: 49: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 49: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0.02] Epoch: 50: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 50: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0.02] Epoch: 51: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 51: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0.02] Epoch: 52: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 52: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0.02] Epoch: 53: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 53: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.02] Epoch: 54: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 54: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0.02] Epoch: 55: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 55: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0.02] Epoch: 56: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 56: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0.01] Epoch: 57: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 57: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0.01] Epoch: 58: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 58: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0.01] Epoch: 59: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 59: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0.01] Epoch: 60: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 60: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0.01] Epoch: 61: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 61: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0.01] Epoch: 62: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 62: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0.01] Epoch: 63: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 63: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0.01] Epoch: 64: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 64: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0.01] Epoch: 65: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 65: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0.01] Epoch: 66: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 66: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0.01] Epoch: 67: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 67: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0.01] Epoch: 68: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 68: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0.01] Epoch: 69: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 69: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0.01] Epoch: 70: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 70: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0.01] Epoch: 71: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 71: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0.01] Epoch: 72: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 72: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0.01] Epoch: 73: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 73: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0.01] Epoch: 74: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 74: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0.01] Epoch: 75: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 75: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0.01] Epoch: 76: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 76: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0.01] Epoch: 77: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 77: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0.01] Epoch: 78: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 78: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0.01] Epoch: 79: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 79: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0.01] Epoch: 80: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 80: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0.01] Epoch: 81: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 81: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0.01] Epoch: 82: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 82: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0.01] Epoch: 83: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 83: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0.01] Epoch: 84: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 84: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 85: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 85: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 86: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 86: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 87: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 87: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0.01] Epoch: 88: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 88: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0.01] Epoch: 89: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 89: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 90: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 90: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0.01] Epoch: 91: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 91: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 92: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 92: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 93: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 93: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 94: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 94: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 95: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 95: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 96: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 96: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0.02] Epoch: 97: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 97: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0.01] Epoch: 98: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 98: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 99: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 99: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 100: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 100: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0.01] Epoch: 101: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 101: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 102: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 102: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0.01] Epoch: 103: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 103: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 104: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 104: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 105: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 105: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 106: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 106: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 107: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 107: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 108: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 108: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 109: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 109: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 110: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 110: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 111: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 111: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 112: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 112: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 113: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 113: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 114: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 114: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 115: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 115: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 116: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 116: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 117: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 117: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 118: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 118: 100% 213/213 [00:10<00:00, 20.06it/s, loss=0] Epoch: 119: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 119: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 120: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 120: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 121: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 121: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 122: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 122: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0.01] Epoch: 123: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 123: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 124: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 124: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 125: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 125: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 126: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 126: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 127: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 127: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 128: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 128: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 129: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 129: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 130: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 130: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 131: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 131: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 132: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 132: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 133: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 133: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 134: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 134: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 135: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 135: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 136: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 136: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 137: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 137: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 138: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 138: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 139: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 139: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 140: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 140: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 141: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 141: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 142: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 142: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 143: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 143: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 144: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 144: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 145: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 145: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 146: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 146: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 147: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 147: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 148: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 148: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 149: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 149: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 150: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 150: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 151: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 151: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 152: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 152: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 153: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 153: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 154: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 154: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 155: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 155: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 156: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 156: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 157: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 157: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 158: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 158: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 159: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 159: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 160: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 160: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 161: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 161: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 162: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 162: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 163: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 163: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 164: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 164: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 165: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 165: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 166: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 166: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 167: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 167: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 168: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 168: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 169: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 169: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 170: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 170: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 171: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 171: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 172: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 172: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 173: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 173: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 174: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 174: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 175: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 175: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 176: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 176: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 177: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 177: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 178: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 178: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 179: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 179: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 180: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 180: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 181: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 181: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 182: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 182: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 183: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 183: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 184: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 184: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 185: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 185: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 186: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 186: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 187: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 187: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 188: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 188: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 189: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 189: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 190: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 190: 100% 213/213 [00:10<00:00, 20.13it/s, loss=0] Epoch: 191: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 191: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 192: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 192: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 193: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 193: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 194: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 194: 100% 213/213 [00:10<00:00, 20.09it/s, loss=0] Epoch: 195: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 195: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 196: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 196: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 197: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 197: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 198: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 198: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 199: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 199: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 200: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 200: 100% 213/213 [00:10<00:00, 20.23it/s, loss=0] Epoch: 201: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 201: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 202: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 202: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 203: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 203: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 204: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 204: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 205: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 205: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 206: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 206: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 207: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 207: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 208: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 208: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 209: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 209: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 210: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 210: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 211: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 211: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 212: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 212: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 213: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 213: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 214: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 214: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 215: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 215: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 216: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 216: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 217: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 217: 100% 213/213 [00:10<00:00, 20.22it/s, loss=0] Epoch: 218: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 218: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 219: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 219: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 220: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 220: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 221: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 221: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 222: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 222: 100% 213/213 [00:10<00:00, 20.07it/s, loss=0] Epoch: 223: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 223: 100% 213/213 [00:10<00:00, 20.05it/s, loss=0] Epoch: 224: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 224: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 225: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 225: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 226: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 226: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 227: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 227: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 228: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 228: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 229: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 229: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 230: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 230: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 231: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 231: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 232: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 232: 100% 213/213 [00:10<00:00, 20.12it/s, loss=0] Epoch: 233: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 233: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 234: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 234: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 235: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 235: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 236: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 236: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 237: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 237: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 238: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 238: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 239: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 239: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 240: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 240: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 241: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 241: 100% 213/213 [00:10<00:00, 20.24it/s, loss=0] Epoch: 242: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 242: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 243: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 243: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 244: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 244: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 245: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 245: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 246: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 246: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 247: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 247: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 248: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 248: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 249: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 249: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 250: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 250: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 251: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 251: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 252: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 252: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 253: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 253: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 254: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 254: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 255: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 255: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 256: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 256: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 257: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 257: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 258: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 258: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 259: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 259: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 260: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 260: 100% 213/213 [00:10<00:00, 20.20it/s, loss=0] Epoch: 261: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 261: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 262: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 262: 100% 213/213 [00:10<00:00, 20.08it/s, loss=0] Epoch: 263: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 263: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 264: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 264: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 265: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 265: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 266: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 266: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 267: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 267: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 268: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 268: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 269: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 269: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 270: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 270: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 271: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 271: 100% 213/213 [00:10<00:00, 20.21it/s, loss=0] Epoch: 272: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 272: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 273: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 273: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 274: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 274: 100% 213/213 [00:10<00:00, 20.03it/s, loss=0] Epoch: 275: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 275: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 276: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 276: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 277: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 277: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0] Epoch: 278: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 278: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 279: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 279: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 280: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 280: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 281: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 281: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 282: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 282: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0] Epoch: 283: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 283: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 284: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 284: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 285: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 285: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 286: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 286: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 287: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 287: 100% 213/213 [00:10<00:00, 20.14it/s, loss=0] Epoch: 288: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 288: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 289: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 289: 100% 213/213 [00:10<00:00, 20.19it/s, loss=0] Epoch: 290: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 290: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 291: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 291: 100% 213/213 [00:10<00:00, 20.10it/s, loss=0] Epoch: 292: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 292: 100% 213/213 [00:10<00:00, 20.16it/s, loss=0] Epoch: 293: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 293: 100% 213/213 [00:10<00:00, 20.18it/s, loss=0] Epoch: 294: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 294: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 295: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 295: 100% 213/213 [00:10<00:00, 20.17it/s, loss=0] Epoch: 296: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 296: 100% 213/213 [00:10<00:00, 20.11it/s, loss=0] Epoch: 297: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 297: 100% 213/213 [00:10<00:00, 20.15it/s, loss=0] Epoch: 298: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 298: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 299: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 299: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0] Epoch: 100% 300/300 [53:15<00:00, 10.65s/it] INFO:root:Training is finished ... Iteration: 100% 24/24 [00:10<00:00, 2.24it/s, acc=1] Iteration: 100% 165/165 [19:26<00:00, 7.07s/it, acc=0.888] obj_pp_to_subj_pp: 21.7 cp_recursion: 52.8 pp_recursion: 37.8 subj_to_obj_proper: 98.0 prim_to_obj_proper: 97.4 prim_to_subj_proper: 100.0 LEX: 97.18 OVERALL: 88.82857142857142 INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 4344077 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.91 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:10<00:00, 19.48it/s, loss=6.06] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:10<00:00, 19.70it/s, loss=4.68] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:10<00:00, 19.78it/s, loss=3.69] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:10<00:00, 19.79it/s, loss=2.59] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:10<00:00, 19.79it/s, loss=2] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:10<00:00, 19.79it/s, loss=1.67] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:10<00:00, 19.77it/s, loss=1.43] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:10<00:00, 19.71it/s, loss=1.26] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:10<00:00, 19.86it/s, loss=1.15] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:10<00:00, 19.94it/s, loss=1.07] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0.98] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0.9] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0.83] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0.73] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0.65] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0.57] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0.49] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0.41] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0.36] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:10<00:00, 20.01it/s, loss=0.3] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0.27] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0.22] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0.2] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0.17] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:10<00:00, 19.68it/s, loss=0.16] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0.17] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0.12] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0.12] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0.1] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0.1] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:10<00:00, 19.71it/s, loss=0.08] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0.08] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0.06] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0.07] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0.06] Epoch: 35: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 35: 100% 213/213 [00:10<00:00, 19.71it/s, loss=0.05] Epoch: 36: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 36: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0.06] Epoch: 37: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 37: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0.05] Epoch: 38: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 38: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0.04] Epoch: 39: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 39: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0.04] Epoch: 40: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 40: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0.04] Epoch: 41: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 41: 100% 213/213 [00:10<00:00, 19.71it/s, loss=0.04] Epoch: 42: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 42: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0.03] Epoch: 43: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 43: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0.03] Epoch: 44: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 44: 100% 213/213 [00:10<00:00, 19.73it/s, loss=0.03] Epoch: 45: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 45: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0.03] Epoch: 46: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 46: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0.03] Epoch: 47: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 47: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0.03] Epoch: 48: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 48: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0.03] Epoch: 49: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 49: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0.03] Epoch: 50: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 50: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0.02] Epoch: 51: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 51: 100% 213/213 [00:10<00:00, 19.73it/s, loss=0.02] Epoch: 52: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 52: 100% 213/213 [00:10<00:00, 19.69it/s, loss=0.02] Epoch: 53: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 53: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0.02] Epoch: 54: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 54: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0.02] Epoch: 55: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 55: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0.03] Epoch: 56: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 56: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0.01] Epoch: 57: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 57: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0.01] Epoch: 58: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 58: 100% 213/213 [00:10<00:00, 19.70it/s, loss=0.01] Epoch: 59: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 59: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0.01] Epoch: 60: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 60: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0.01] Epoch: 61: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 61: 100% 213/213 [00:10<00:00, 19.72it/s, loss=0.01] Epoch: 62: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 62: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0.01] Epoch: 63: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 63: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0.02] Epoch: 64: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 64: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0.02] Epoch: 65: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 65: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0.01] Epoch: 66: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 66: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0.01] Epoch: 67: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 67: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0.01] Epoch: 68: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 68: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0.01] Epoch: 69: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 69: 100% 213/213 [00:10<00:00, 19.72it/s, loss=0.01] Epoch: 70: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 70: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0.01] Epoch: 71: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 71: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0.01] Epoch: 72: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 72: 100% 213/213 [00:10<00:00, 19.72it/s, loss=0.02] Epoch: 73: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 73: 100% 213/213 [00:10<00:00, 19.71it/s, loss=0.01] Epoch: 74: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 74: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0.01] Epoch: 75: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 75: 100% 213/213 [00:10<00:00, 19.70it/s, loss=0.01] Epoch: 76: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 76: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0.01] Epoch: 77: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 77: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0.02] Epoch: 78: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 78: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0.01] Epoch: 79: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 79: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0.01] Epoch: 80: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 80: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0.01] Epoch: 81: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 81: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0.01] Epoch: 82: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 82: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 83: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 83: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0.01] Epoch: 84: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 84: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0.01] Epoch: 85: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 85: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0.01] Epoch: 86: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 86: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0.01] Epoch: 87: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 87: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0] Epoch: 88: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 88: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0.01] Epoch: 89: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 89: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0.01] Epoch: 90: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 90: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0.01] Epoch: 91: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 91: 100% 213/213 [00:10<00:00, 20.00it/s, loss=0.01] Epoch: 92: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 92: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0.01] Epoch: 93: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 93: 100% 213/213 [00:10<00:00, 20.04it/s, loss=0.01] Epoch: 94: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 94: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0.01] Epoch: 95: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 95: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0.01] Epoch: 96: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 96: 100% 213/213 [00:10<00:00, 20.02it/s, loss=0.01] Epoch: 97: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 97: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 98: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 98: 100% 213/213 [00:10<00:00, 19.95it/s, loss=0] Epoch: 99: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 99: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0.01] Epoch: 100: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 100: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 101: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 101: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0] Epoch: 102: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 102: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0.01] Epoch: 103: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 103: 100% 213/213 [00:10<00:00, 19.70it/s, loss=0] Epoch: 104: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 104: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0] Epoch: 105: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 105: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0] Epoch: 106: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 106: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0.01] Epoch: 107: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 107: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 108: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 108: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0] Epoch: 109: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 109: 100% 213/213 [00:10<00:00, 19.72it/s, loss=0] Epoch: 110: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 110: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0.01] Epoch: 111: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 111: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 112: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 112: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0] Epoch: 113: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 113: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 114: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 114: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0] Epoch: 115: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 115: 100% 213/213 [00:10<00:00, 19.73it/s, loss=0] Epoch: 116: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 116: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 117: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 117: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0.01] Epoch: 118: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 118: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 119: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 119: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0] Epoch: 120: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 120: 100% 213/213 [00:10<00:00, 19.68it/s, loss=0] Epoch: 121: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 121: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0.01] Epoch: 122: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 122: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 123: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 123: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0.01] Epoch: 124: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 124: 100% 213/213 [00:10<00:00, 19.72it/s, loss=0] Epoch: 125: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 125: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0] Epoch: 126: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 126: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0] Epoch: 127: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 127: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 128: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 128: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 129: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 129: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 130: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 130: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0] Epoch: 131: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 131: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0.01] Epoch: 132: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 132: 100% 213/213 [00:10<00:00, 19.72it/s, loss=0] Epoch: 133: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 133: 100% 213/213 [00:10<00:00, 19.72it/s, loss=0] Epoch: 134: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 134: 100% 213/213 [00:10<00:00, 19.70it/s, loss=0] Epoch: 135: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 135: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 136: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 136: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0] Epoch: 137: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 137: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0] Epoch: 138: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 138: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0] Epoch: 139: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 139: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0] Epoch: 140: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 140: 100% 213/213 [00:10<00:00, 19.95it/s, loss=0] Epoch: 141: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 141: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 142: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 142: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 143: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 143: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0] Epoch: 144: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 144: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 145: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 145: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 146: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 146: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0] Epoch: 147: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 147: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0] Epoch: 148: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 148: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 149: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 149: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 150: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 150: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 151: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 151: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 152: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 152: 100% 213/213 [00:10<00:00, 19.95it/s, loss=0] Epoch: 153: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 153: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 154: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 154: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 155: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 155: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 156: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 156: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0] Epoch: 157: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 157: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0] Epoch: 158: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 158: 100% 213/213 [00:10<00:00, 19.73it/s, loss=0] Epoch: 159: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 159: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0] Epoch: 160: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 160: 100% 213/213 [00:10<00:00, 19.69it/s, loss=0] Epoch: 161: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 161: 100% 213/213 [00:10<00:00, 19.73it/s, loss=0] Epoch: 162: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 162: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0] Epoch: 163: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 163: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 164: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 164: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 165: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 165: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0] Epoch: 166: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 166: 100% 213/213 [00:10<00:00, 19.70it/s, loss=0] Epoch: 167: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 167: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 168: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 168: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0] Epoch: 169: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 169: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0] Epoch: 170: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 170: 100% 213/213 [00:10<00:00, 19.73it/s, loss=0] Epoch: 171: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 171: 100% 213/213 [00:10<00:00, 19.67it/s, loss=0] Epoch: 172: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 172: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0] Epoch: 173: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 173: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0] Epoch: 174: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 174: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0] Epoch: 175: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 175: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0] Epoch: 176: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 176: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0] Epoch: 177: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 177: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0] Epoch: 178: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 178: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0] Epoch: 179: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 179: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0] Epoch: 180: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 180: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 181: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 181: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0] Epoch: 182: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 182: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0] Epoch: 183: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 183: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0] Epoch: 184: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 184: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 185: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 185: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0] Epoch: 186: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 186: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0] Epoch: 187: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 187: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0] Epoch: 188: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 188: 100% 213/213 [00:10<00:00, 19.70it/s, loss=0] Epoch: 189: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 189: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 190: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 190: 100% 213/213 [00:10<00:00, 19.73it/s, loss=0] Epoch: 191: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 191: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0] Epoch: 192: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 192: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0] Epoch: 193: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 193: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0] Epoch: 194: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 194: 100% 213/213 [00:10<00:00, 19.71it/s, loss=0] Epoch: 195: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 195: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0] Epoch: 196: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 196: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 197: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 197: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0] Epoch: 198: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 198: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0] Epoch: 199: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 199: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0] Epoch: 200: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 200: 100% 213/213 [00:10<00:00, 19.70it/s, loss=0] Epoch: 201: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 201: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0] Epoch: 202: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 202: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0] Epoch: 203: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 203: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 204: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 204: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 205: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 205: 100% 213/213 [00:10<00:00, 19.73it/s, loss=0] Epoch: 206: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 206: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0] Epoch: 207: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 207: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 208: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 208: 100% 213/213 [00:10<00:00, 19.95it/s, loss=0] Epoch: 209: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 209: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 210: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 210: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 211: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 211: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 212: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 212: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0] Epoch: 213: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 213: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0] Epoch: 214: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 214: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0] Epoch: 215: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 215: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 216: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 216: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0] Epoch: 217: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 217: 100% 213/213 [00:10<00:00, 19.73it/s, loss=0] Epoch: 218: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 218: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 219: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 219: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0] Epoch: 220: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 220: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0] Epoch: 221: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 221: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 222: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 222: 100% 213/213 [00:10<00:00, 19.69it/s, loss=0] Epoch: 223: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 223: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0] Epoch: 224: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 224: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0] Epoch: 225: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 225: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0] Epoch: 226: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 226: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0] Epoch: 227: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 227: 100% 213/213 [00:10<00:00, 19.73it/s, loss=0] Epoch: 228: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 228: 100% 213/213 [00:10<00:00, 19.66it/s, loss=0] Epoch: 229: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 229: 100% 213/213 [00:10<00:00, 19.71it/s, loss=0] Epoch: 230: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 230: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0] Epoch: 231: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 231: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0] Epoch: 232: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 232: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0] Epoch: 233: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 233: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0] Epoch: 234: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 234: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0] Epoch: 235: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 235: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 236: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 236: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 237: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 237: 100% 213/213 [00:10<00:00, 19.81it/s, loss=0] Epoch: 238: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 238: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0] Epoch: 239: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 239: 100% 213/213 [00:10<00:00, 19.70it/s, loss=0] Epoch: 240: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 240: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0] Epoch: 241: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 241: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 242: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 242: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 243: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 243: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 244: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 244: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 245: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 245: 100% 213/213 [00:10<00:00, 19.71it/s, loss=0] Epoch: 246: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 246: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0] Epoch: 247: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 247: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 248: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 248: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0] Epoch: 249: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 249: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0] Epoch: 250: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 250: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 251: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 251: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0] Epoch: 252: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 252: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0] Epoch: 253: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 253: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0] Epoch: 254: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 254: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 255: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 255: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 256: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 256: 100% 213/213 [00:10<00:00, 19.65it/s, loss=0] Epoch: 257: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 257: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0] Epoch: 258: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 258: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0] Epoch: 259: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 259: 100% 213/213 [00:10<00:00, 19.72it/s, loss=0] Epoch: 260: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 260: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0] Epoch: 261: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 261: 100% 213/213 [00:10<00:00, 19.72it/s, loss=0] Epoch: 262: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 262: 100% 213/213 [00:10<00:00, 19.71it/s, loss=0] Epoch: 263: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 263: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0] Epoch: 264: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 264: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0] Epoch: 265: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 265: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 266: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 266: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 267: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 267: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0] Epoch: 268: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 268: 100% 213/213 [00:10<00:00, 19.71it/s, loss=0] Epoch: 269: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 269: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 270: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 270: 100% 213/213 [00:10<00:00, 19.71it/s, loss=0] Epoch: 271: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 271: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 272: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 272: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0] Epoch: 273: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 273: 100% 213/213 [00:10<00:00, 19.68it/s, loss=0] Epoch: 274: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 274: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0] Epoch: 275: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 275: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0] Epoch: 276: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 276: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0] Epoch: 277: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 277: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 278: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 278: 100% 213/213 [00:10<00:00, 19.95it/s, loss=0] Epoch: 279: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 279: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0] Epoch: 280: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 280: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 281: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 281: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 282: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 282: 100% 213/213 [00:10<00:00, 19.94it/s, loss=0] Epoch: 283: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 283: 100% 213/213 [00:10<00:00, 19.98it/s, loss=0] Epoch: 284: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 284: 100% 213/213 [00:10<00:00, 19.99it/s, loss=0] Epoch: 285: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 285: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 286: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 286: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 287: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 287: 100% 213/213 [00:10<00:00, 19.97it/s, loss=0] Epoch: 288: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 288: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 289: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 289: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0] Epoch: 290: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 290: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0] Epoch: 291: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 291: 100% 213/213 [00:10<00:00, 19.66it/s, loss=0] Epoch: 292: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 292: 100% 213/213 [00:10<00:00, 19.72it/s, loss=0] Epoch: 293: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 293: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0] Epoch: 294: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 294: 100% 213/213 [00:10<00:00, 19.73it/s, loss=0] Epoch: 295: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 295: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0] Epoch: 296: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 296: 100% 213/213 [00:10<00:00, 19.73it/s, loss=0] Epoch: 297: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 297: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0] Epoch: 298: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 298: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 299: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 299: 100% 213/213 [00:10<00:00, 19.72it/s, loss=0] Epoch: 100% 300/300 [54:08<00:00, 10.83s/it] INFO:root:Training is finished ... Iteration: 100% 24/24 [00:10<00:00, 2.26it/s, acc=1] Iteration: 100% 165/165 [18:44<00:00, 6.82s/it, acc=0.854] obj_pp_to_subj_pp: 18.9 cp_recursion: 53.3 pp_recursion: 46.9 subj_to_obj_proper: 86.8 prim_to_obj_proper: 72.5 prim_to_subj_proper: 99.9 LEX: 94.36666666666666 OVERALL: 85.41904761904762 INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 4344077 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.102 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:10<00:00, 19.44it/s, loss=5.99] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:10<00:00, 19.68it/s, loss=4.57] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:10<00:00, 19.67it/s, loss=3.44] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:10<00:00, 19.72it/s, loss=2.41] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:10<00:00, 19.67it/s, loss=1.88] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:10<00:00, 19.65it/s, loss=1.57] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:10<00:00, 19.70it/s, loss=1.3] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:10<00:00, 19.71it/s, loss=1.11] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:10<00:00, 19.69it/s, loss=0.98] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0.86] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:10<00:00, 19.65it/s, loss=0.78] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:10<00:00, 19.68it/s, loss=0.69] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:10<00:00, 19.69it/s, loss=0.62] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:10<00:00, 19.67it/s, loss=0.54] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:10<00:00, 19.72it/s, loss=0.48] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:10<00:00, 19.68it/s, loss=0.41] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:10<00:00, 19.59it/s, loss=0.36] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:10<00:00, 19.65it/s, loss=0.31] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:10<00:00, 19.66it/s, loss=0.26] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:10<00:00, 19.69it/s, loss=0.21] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0.18] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:10<00:00, 19.72it/s, loss=0.16] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0.13] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0.11] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0.1] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0.08] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0.08] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0.07] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0.06] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0.05] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0.05] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0.04] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:10<00:00, 19.92it/s, loss=0.04] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0.03] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0.03] Epoch: 35: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 35: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0.02] Epoch: 36: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 36: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0.03] Epoch: 37: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 37: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0.02] Epoch: 38: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 38: 100% 213/213 [00:10<00:00, 19.82it/s, loss=0.02] Epoch: 39: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 39: 100% 213/213 [00:10<00:00, 19.60it/s, loss=0.02] Epoch: 40: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 40: 100% 213/213 [00:10<00:00, 19.66it/s, loss=0.02] Epoch: 41: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 41: 100% 213/213 [00:10<00:00, 19.69it/s, loss=0.03] Epoch: 42: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 42: 100% 213/213 [00:10<00:00, 19.69it/s, loss=0.01] Epoch: 43: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 43: 100% 213/213 [00:10<00:00, 19.64it/s, loss=0.02] Epoch: 44: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 44: 100% 213/213 [00:10<00:00, 19.60it/s, loss=0.01] Epoch: 45: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 45: 100% 213/213 [00:10<00:00, 19.68it/s, loss=0.01] Epoch: 46: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 46: 100% 213/213 [00:10<00:00, 19.72it/s, loss=0.01] Epoch: 47: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 47: 100% 213/213 [00:10<00:00, 19.67it/s, loss=0.01] Epoch: 48: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 48: 100% 213/213 [00:10<00:00, 19.62it/s, loss=0.01] Epoch: 49: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 49: 100% 213/213 [00:10<00:00, 19.67it/s, loss=0.01] Epoch: 50: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 50: 100% 213/213 [00:10<00:00, 19.64it/s, loss=0.01] Epoch: 51: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 51: 100% 213/213 [00:10<00:00, 19.70it/s, loss=0.01] Epoch: 52: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 52: 100% 213/213 [00:10<00:00, 19.72it/s, loss=0.01] Epoch: 53: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 53: 100% 213/213 [00:10<00:00, 19.70it/s, loss=0.01] Epoch: 54: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 54: 100% 213/213 [00:10<00:00, 19.68it/s, loss=0.01] Epoch: 55: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 55: 100% 213/213 [00:10<00:00, 19.68it/s, loss=0.01] Epoch: 56: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 56: 100% 213/213 [00:10<00:00, 19.60it/s, loss=0.01] Epoch: 57: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 57: 100% 213/213 [00:10<00:00, 19.67it/s, loss=0.01] Epoch: 58: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 58: 100% 213/213 [00:10<00:00, 19.71it/s, loss=0.01] Epoch: 59: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 59: 100% 213/213 [00:10<00:00, 19.69it/s, loss=0.01] Epoch: 60: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 60: 100% 213/213 [00:10<00:00, 19.71it/s, loss=0] Epoch: 61: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 61: 100% 213/213 [00:10<00:00, 19.62it/s, loss=0] Epoch: 62: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 62: 100% 213/213 [00:10<00:00, 19.68it/s, loss=0.01] Epoch: 63: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 63: 100% 213/213 [00:10<00:00, 19.69it/s, loss=0.01] Epoch: 64: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 64: 100% 213/213 [00:10<00:00, 19.69it/s, loss=0] Epoch: 65: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 65: 100% 213/213 [00:10<00:00, 19.66it/s, loss=0.01] Epoch: 66: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 66: 100% 213/213 [00:10<00:00, 19.66it/s, loss=0] Epoch: 67: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 67: 100% 213/213 [00:10<00:00, 19.65it/s, loss=0.01] Epoch: 68: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 68: 100% 213/213 [00:10<00:00, 19.68it/s, loss=0] Epoch: 69: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 69: 100% 213/213 [00:10<00:00, 19.70it/s, loss=0] Epoch: 70: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 70: 100% 213/213 [00:10<00:00, 19.72it/s, loss=0] Epoch: 71: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 71: 100% 213/213 [00:10<00:00, 19.67it/s, loss=0] Epoch: 72: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 72: 100% 213/213 [00:10<00:00, 19.65it/s, loss=0] Epoch: 73: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 73: 100% 213/213 [00:10<00:00, 19.66it/s, loss=0] Epoch: 74: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 74: 100% 213/213 [00:10<00:00, 19.72it/s, loss=0] Epoch: 75: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 75: 100% 213/213 [00:10<00:00, 19.70it/s, loss=0.01] Epoch: 76: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 76: 100% 213/213 [00:10<00:00, 19.70it/s, loss=0] Epoch: 77: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 77: 100% 213/213 [00:10<00:00, 19.68it/s, loss=0] Epoch: 78: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 78: 100% 213/213 [00:10<00:00, 19.62it/s, loss=0] Epoch: 79: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 79: 100% 213/213 [00:10<00:00, 19.70it/s, loss=0.01] Epoch: 80: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 80: 100% 213/213 [00:10<00:00, 19.71it/s, loss=0] Epoch: 81: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 81: 100% 213/213 [00:10<00:00, 19.70it/s, loss=0] Epoch: 82: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 82: 100% 213/213 [00:10<00:00, 19.73it/s, loss=0] Epoch: 83: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 83: 100% 213/213 [00:10<00:00, 19.70it/s, loss=0.01] Epoch: 84: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 84: 100% 213/213 [00:10<00:00, 19.65it/s, loss=0] Epoch: 85: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 85: 100% 213/213 [00:10<00:00, 19.71it/s, loss=0] Epoch: 86: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 86: 100% 213/213 [00:10<00:00, 19.73it/s, loss=0] Epoch: 87: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 87: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0] Epoch: 88: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 88: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 89: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 89: 100% 213/213 [00:10<00:00, 19.90it/s, loss=0] Epoch: 90: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 90: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 91: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 91: 100% 213/213 [00:10<00:00, 19.93it/s, loss=0] Epoch: 92: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 92: 100% 213/213 [00:10<00:00, 19.96it/s, loss=0] Epoch: 93: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 93: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 94: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 94: 100% 213/213 [00:10<00:00, 19.91it/s, loss=0] Epoch: 95: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 95: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0] Epoch: 96: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 96: 100% 213/213 [00:10<00:00, 19.68it/s, loss=0] Epoch: 97: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 97: 100% 213/213 [00:10<00:00, 19.68it/s, loss=0] Epoch: 98: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 98: 100% 213/213 [00:10<00:00, 19.71it/s, loss=0] Epoch: 99: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 99: 100% 213/213 [00:10<00:00, 19.69it/s, loss=0] Epoch: 100: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 100: 100% 213/213 [00:10<00:00, 19.69it/s, loss=0] Epoch: 101: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 101: 100% 213/213 [00:10<00:00, 19.64it/s, loss=0] Epoch: 102: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 102: 100% 213/213 [00:10<00:00, 19.71it/s, loss=0] Epoch: 103: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 103: 100% 213/213 [00:10<00:00, 19.68it/s, loss=0] Epoch: 104: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 104: 100% 213/213 [00:10<00:00, 19.66it/s, loss=0] Epoch: 105: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 105: 100% 213/213 [00:10<00:00, 19.68it/s, loss=0] Epoch: 106: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 106: 100% 213/213 [00:10<00:00, 19.70it/s, loss=0] Epoch: 107: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 107: 100% 213/213 [00:10<00:00, 19.64it/s, loss=0] Epoch: 108: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 108: 100% 213/213 [00:10<00:00, 19.73it/s, loss=0] Epoch: 109: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 109: 100% 213/213 [00:10<00:00, 19.69it/s, loss=0] Epoch: 110: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 110: 100% 213/213 [00:10<00:00, 19.72it/s, loss=0] Epoch: 111: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 111: 100% 213/213 [00:10<00:00, 19.64it/s, loss=0] Epoch: 112: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 112: 100% 213/213 [00:10<00:00, 19.61it/s, loss=0] Epoch: 113: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 113: 100% 213/213 [00:10<00:00, 19.70it/s, loss=0] Epoch: 114: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 114: 100% 213/213 [00:10<00:00, 19.69it/s, loss=0] Epoch: 115: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 115: 100% 213/213 [00:10<00:00, 19.71it/s, loss=0] Epoch: 116: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 116: 100% 213/213 [00:10<00:00, 19.65it/s, loss=0] Epoch: 117: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 117: 100% 213/213 [00:10<00:00, 19.67it/s, loss=0] Epoch: 118: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 118: 100% 213/213 [00:10<00:00, 19.61it/s, loss=0] Epoch: 119: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 119: 100% 213/213 [00:10<00:00, 19.70it/s, loss=0] Epoch: 120: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 120: 100% 213/213 [00:10<00:00, 19.70it/s, loss=0] Epoch: 121: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 121: 100% 213/213 [00:10<00:00, 19.68it/s, loss=0] Epoch: 122: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 122: 100% 213/213 [00:10<00:00, 19.65it/s, loss=0] Epoch: 123: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 123: 100% 213/213 [00:10<00:00, 19.66it/s, loss=0] Epoch: 124: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 124: 100% 213/213 [00:10<00:00, 19.67it/s, loss=0] Epoch: 125: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 125: 100% 213/213 [00:10<00:00, 19.64it/s, loss=0] Epoch: 126: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 126: 100% 213/213 [00:10<00:00, 19.68it/s, loss=0] Epoch: 127: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 127: 100% 213/213 [00:10<00:00, 19.65it/s, loss=0] Epoch: 128: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 128: 100% 213/213 [00:10<00:00, 19.70it/s, loss=0] Epoch: 129: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 129: 100% 213/213 [00:10<00:00, 19.59it/s, loss=0] Epoch: 130: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 130: 100% 213/213 [00:10<00:00, 19.67it/s, loss=0] Epoch: 131: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 131: 100% 213/213 [00:10<00:00, 19.70it/s, loss=0] Epoch: 132: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 132: 100% 213/213 [00:10<00:00, 19.70it/s, loss=0] Epoch: 133: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 133: 100% 213/213 [00:10<00:00, 19.64it/s, loss=0] Epoch: 134: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 134: 100% 213/213 [00:10<00:00, 19.71it/s, loss=0] Epoch: 135: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 135: 100% 213/213 [00:10<00:00, 19.62it/s, loss=0] Epoch: 136: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 136: 100% 213/213 [00:10<00:00, 19.71it/s, loss=0] Epoch: 137: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 137: 100% 213/213 [00:10<00:00, 19.70it/s, loss=0] Epoch: 138: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 138: 100% 213/213 [00:10<00:00, 19.69it/s, loss=0] Epoch: 139: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 139: 100% 213/213 [00:10<00:00, 19.66it/s, loss=0] Epoch: 140: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 140: 100% 213/213 [00:10<00:00, 19.55it/s, loss=0] Epoch: 141: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 141: 100% 213/213 [00:10<00:00, 19.66it/s, loss=0] Epoch: 142: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 142: 100% 213/213 [00:10<00:00, 19.71it/s, loss=0] Epoch: 143: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 143: 100% 213/213 [00:10<00:00, 19.68it/s, loss=0] Epoch: 144: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 144: 100% 213/213 [00:10<00:00, 19.72it/s, loss=0] Epoch: 145: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 145: 100% 213/213 [00:10<00:00, 19.69it/s, loss=0] Epoch: 146: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 146: 100% 213/213 [00:10<00:00, 19.65it/s, loss=0] Epoch: 147: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 147: 100% 213/213 [00:10<00:00, 19.72it/s, loss=0] Epoch: 148: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 148: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0] Epoch: 149: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 149: 100% 213/213 [00:10<00:00, 19.69it/s, loss=0] Epoch: 150: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 150: 100% 213/213 [00:10<00:00, 19.70it/s, loss=0] Epoch: 151: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 151: 100% 213/213 [00:10<00:00, 19.71it/s, loss=0] Epoch: 152: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 152: 100% 213/213 [00:10<00:00, 19.67it/s, loss=0] Epoch: 153: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 153: 100% 213/213 [00:10<00:00, 19.71it/s, loss=0] Epoch: 154: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 154: 100% 213/213 [00:10<00:00, 19.72it/s, loss=0] Epoch: 155: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 155: 100% 213/213 [00:10<00:00, 19.69it/s, loss=0] Epoch: 156: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 156: 100% 213/213 [00:10<00:00, 19.67it/s, loss=0] Epoch: 157: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 157: 100% 213/213 [00:10<00:00, 19.68it/s, loss=0] Epoch: 158: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 158: 100% 213/213 [00:10<00:00, 19.68it/s, loss=0] Epoch: 159: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 159: 100% 213/213 [00:10<00:00, 19.68it/s, loss=0] Epoch: 160: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 160: 100% 213/213 [00:10<00:00, 19.69it/s, loss=0] Epoch: 161: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 161: 100% 213/213 [00:10<00:00, 19.69it/s, loss=0] Epoch: 162: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 162: 100% 213/213 [00:10<00:00, 19.68it/s, loss=0] Epoch: 163: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 163: 100% 213/213 [00:10<00:00, 19.60it/s, loss=0] Epoch: 164: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 164: 100% 213/213 [00:10<00:00, 19.63it/s, loss=0] Epoch: 165: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 165: 100% 213/213 [00:10<00:00, 19.67it/s, loss=0] Epoch: 166: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 166: 100% 213/213 [00:10<00:00, 19.66it/s, loss=0] Epoch: 167: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 167: 100% 213/213 [00:10<00:00, 19.65it/s, loss=0] Epoch: 168: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 168: 100% 213/213 [00:10<00:00, 19.64it/s, loss=0] Epoch: 169: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 169: 100% 213/213 [00:10<00:00, 19.65it/s, loss=0] Epoch: 170: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 170: 100% 213/213 [00:10<00:00, 19.67it/s, loss=0] Epoch: 171: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 171: 100% 213/213 [00:10<00:00, 19.70it/s, loss=0] Epoch: 172: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 172: 100% 213/213 [00:10<00:00, 19.69it/s, loss=0] Epoch: 173: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 173: 100% 213/213 [00:10<00:00, 19.66it/s, loss=0] Epoch: 174: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 174: 100% 213/213 [00:10<00:00, 19.63it/s, loss=0] Epoch: 175: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 175: 100% 213/213 [00:10<00:00, 19.66it/s, loss=0] Epoch: 176: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 176: 100% 213/213 [00:10<00:00, 19.65it/s, loss=0] Epoch: 177: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 177: 100% 213/213 [00:10<00:00, 19.67it/s, loss=0] Epoch: 178: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 178: 100% 213/213 [00:10<00:00, 19.64it/s, loss=0] Epoch: 179: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 179: 100% 213/213 [00:10<00:00, 19.64it/s, loss=0] Epoch: 180: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 180: 100% 213/213 [00:10<00:00, 19.61it/s, loss=0] Epoch: 181: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 181: 100% 213/213 [00:10<00:00, 19.72it/s, loss=0] Epoch: 182: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 182: 100% 213/213 [00:10<00:00, 19.62it/s, loss=0] Epoch: 183: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 183: 100% 213/213 [00:10<00:00, 19.64it/s, loss=0] Epoch: 184: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 184: 100% 213/213 [00:10<00:00, 19.66it/s, loss=0] Epoch: 185: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 185: 100% 213/213 [00:10<00:00, 19.65it/s, loss=0] Epoch: 186: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 186: 100% 213/213 [00:10<00:00, 19.61it/s, loss=0] Epoch: 187: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 187: 100% 213/213 [00:10<00:00, 19.65it/s, loss=0] Epoch: 188: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 188: 100% 213/213 [00:10<00:00, 19.69it/s, loss=0] Epoch: 189: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 189: 100% 213/213 [00:10<00:00, 19.68it/s, loss=0] Epoch: 190: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 190: 100% 213/213 [00:10<00:00, 19.66it/s, loss=0] Epoch: 191: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 191: 100% 213/213 [00:10<00:00, 19.63it/s, loss=0] Epoch: 192: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 192: 100% 213/213 [00:10<00:00, 19.65it/s, loss=0] Epoch: 193: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 193: 100% 213/213 [00:10<00:00, 19.66it/s, loss=0] Epoch: 194: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 194: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 195: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 195: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 196: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 196: 100% 213/213 [00:10<00:00, 19.88it/s, loss=0] Epoch: 197: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 197: 100% 213/213 [00:10<00:00, 19.78it/s, loss=0] Epoch: 198: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 198: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 199: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 199: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 200: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 200: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 201: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 201: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 202: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 202: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 203: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 203: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 204: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 204: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 205: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 205: 100% 213/213 [00:10<00:00, 19.83it/s, loss=0] Epoch: 206: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 206: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 207: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 207: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 208: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 208: 100% 213/213 [00:10<00:00, 19.76it/s, loss=0] Epoch: 209: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 209: 100% 213/213 [00:10<00:00, 19.86it/s, loss=0] Epoch: 210: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 210: 100% 213/213 [00:10<00:00, 19.89it/s, loss=0] Epoch: 211: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 211: 100% 213/213 [00:10<00:00, 19.84it/s, loss=0] Epoch: 212: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 212: 100% 213/213 [00:10<00:00, 19.79it/s, loss=0] Epoch: 213: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 213: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 214: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 214: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0] Epoch: 215: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 215: 100% 213/213 [00:10<00:00, 19.74it/s, loss=0] Epoch: 216: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 216: 100% 213/213 [00:10<00:00, 19.61it/s, loss=0] Epoch: 217: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 217: 100% 213/213 [00:10<00:00, 19.64it/s, loss=0] Epoch: 218: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 218: 100% 213/213 [00:10<00:00, 19.63it/s, loss=0] Epoch: 219: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 219: 100% 213/213 [00:10<00:00, 19.64it/s, loss=0] Epoch: 220: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 220: 100% 213/213 [00:10<00:00, 19.57it/s, loss=0] Epoch: 221: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 221: 100% 213/213 [00:10<00:00, 19.68it/s, loss=0] Epoch: 222: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 222: 100% 213/213 [00:10<00:00, 19.69it/s, loss=0] Epoch: 223: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 223: 100% 213/213 [00:10<00:00, 19.68it/s, loss=0] Epoch: 224: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 224: 100% 213/213 [00:10<00:00, 19.67it/s, loss=0] Epoch: 225: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 225: 100% 213/213 [00:10<00:00, 19.58it/s, loss=0] Epoch: 226: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 226: 100% 213/213 [00:10<00:00, 19.63it/s, loss=0] Epoch: 227: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 227: 100% 213/213 [00:10<00:00, 19.66it/s, loss=0] Epoch: 228: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 228: 100% 213/213 [00:10<00:00, 19.66it/s, loss=0] Epoch: 229: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 229: 100% 213/213 [00:10<00:00, 19.66it/s, loss=0] Epoch: 230: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 230: 100% 213/213 [00:10<00:00, 19.65it/s, loss=0] Epoch: 231: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 231: 100% 213/213 [00:10<00:00, 19.62it/s, loss=0] Epoch: 232: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 232: 100% 213/213 [00:10<00:00, 19.67it/s, loss=0] Epoch: 233: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 233: 100% 213/213 [00:10<00:00, 19.66it/s, loss=0] Epoch: 234: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 234: 100% 213/213 [00:10<00:00, 19.65it/s, loss=0] Epoch: 235: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 235: 100% 213/213 [00:10<00:00, 19.69it/s, loss=0] Epoch: 236: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 236: 100% 213/213 [00:10<00:00, 19.60it/s, loss=0] Epoch: 237: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 237: 100% 213/213 [00:10<00:00, 19.64it/s, loss=0] Epoch: 238: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 238: 100% 213/213 [00:10<00:00, 19.67it/s, loss=0] Epoch: 239: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 239: 100% 213/213 [00:10<00:00, 19.67it/s, loss=0] Epoch: 240: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 240: 100% 213/213 [00:10<00:00, 19.67it/s, loss=0] Epoch: 241: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 241: 100% 213/213 [00:10<00:00, 19.64it/s, loss=0] Epoch: 242: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 242: 100% 213/213 [00:10<00:00, 19.69it/s, loss=0] Epoch: 243: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 243: 100% 213/213 [00:10<00:00, 19.80it/s, loss=0] Epoch: 244: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 244: 100% 213/213 [00:10<00:00, 19.87it/s, loss=0] Epoch: 245: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 245: 100% 213/213 [00:10<00:00, 19.69it/s, loss=0] Epoch: 246: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 246: 100% 213/213 [00:10<00:00, 19.68it/s, loss=0] Epoch: 247: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 247: 100% 213/213 [00:10<00:00, 19.69it/s, loss=0] Epoch: 248: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 248: 100% 213/213 [00:10<00:00, 19.62it/s, loss=0] Epoch: 249: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 249: 100% 213/213 [00:10<00:00, 19.67it/s, loss=0] Epoch: 250: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 250: 100% 213/213 [00:10<00:00, 19.67it/s, loss=0] Epoch: 251: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 251: 100% 213/213 [00:10<00:00, 19.69it/s, loss=0] Epoch: 252: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 252: 100% 213/213 [00:10<00:00, 19.69it/s, loss=0] Epoch: 253: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 253: 100% 213/213 [00:10<00:00, 19.65it/s, loss=0] Epoch: 254: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 254: 100% 213/213 [00:10<00:00, 19.70it/s, loss=0] Epoch: 255: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 255: 100% 213/213 [00:10<00:00, 19.70it/s, loss=0] Epoch: 256: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 256: 100% 213/213 [00:10<00:00, 19.68it/s, loss=0] Epoch: 257: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 257: 100% 213/213 [00:10<00:00, 19.68it/s, loss=0] Epoch: 258: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 258: 100% 213/213 [00:10<00:00, 19.70it/s, loss=0] Epoch: 259: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 259: 100% 213/213 [00:10<00:00, 19.63it/s, loss=0] Epoch: 260: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 260: 100% 213/213 [00:10<00:00, 19.66it/s, loss=0] Epoch: 261: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 261: 100% 213/213 [00:10<00:00, 19.66it/s, loss=0] Epoch: 262: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 262: 100% 213/213 [00:10<00:00, 19.70it/s, loss=0] Epoch: 263: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 263: 100% 213/213 [00:10<00:00, 19.68it/s, loss=0] Epoch: 264: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 264: 100% 213/213 [00:10<00:00, 19.66it/s, loss=0] Epoch: 265: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 265: 100% 213/213 [00:10<00:00, 19.61it/s, loss=0] Epoch: 266: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 266: 100% 213/213 [00:10<00:00, 19.70it/s, loss=0] Epoch: 267: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 267: 100% 213/213 [00:10<00:00, 19.67it/s, loss=0] Epoch: 268: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 268: 100% 213/213 [00:10<00:00, 19.66it/s, loss=0] Epoch: 269: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 269: 100% 213/213 [00:10<00:00, 19.67it/s, loss=0] Epoch: 270: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 270: 100% 213/213 [00:10<00:00, 19.63it/s, loss=0] Epoch: 271: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 271: 100% 213/213 [00:10<00:00, 19.67it/s, loss=0] Epoch: 272: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 272: 100% 213/213 [00:10<00:00, 19.64it/s, loss=0] Epoch: 273: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 273: 100% 213/213 [00:10<00:00, 19.69it/s, loss=0] Epoch: 274: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 274: 100% 213/213 [00:10<00:00, 19.64it/s, loss=0] Epoch: 275: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 275: 100% 213/213 [00:10<00:00, 19.67it/s, loss=0] Epoch: 276: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 276: 100% 213/213 [00:10<00:00, 19.59it/s, loss=0] Epoch: 277: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 277: 100% 213/213 [00:10<00:00, 19.70it/s, loss=0] Epoch: 278: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 278: 100% 213/213 [00:10<00:00, 19.70it/s, loss=0] Epoch: 279: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 279: 100% 213/213 [00:10<00:00, 19.66it/s, loss=0] Epoch: 280: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 280: 100% 213/213 [00:10<00:00, 19.65it/s, loss=0] Epoch: 281: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 281: 100% 213/213 [00:10<00:00, 19.65it/s, loss=0] Epoch: 282: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 282: 100% 213/213 [00:10<00:00, 19.64it/s, loss=0] Epoch: 283: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 283: 100% 213/213 [00:10<00:00, 19.68it/s, loss=0] Epoch: 284: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 284: 100% 213/213 [00:10<00:00, 19.70it/s, loss=0] Epoch: 285: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 285: 100% 213/213 [00:10<00:00, 19.72it/s, loss=0] Epoch: 286: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 286: 100% 213/213 [00:10<00:00, 19.72it/s, loss=0] Epoch: 287: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 287: 100% 213/213 [00:10<00:00, 19.65it/s, loss=0] Epoch: 288: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 288: 100% 213/213 [00:10<00:00, 19.67it/s, loss=0] Epoch: 289: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 289: 100% 213/213 [00:10<00:00, 19.68it/s, loss=0] Epoch: 290: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 290: 100% 213/213 [00:10<00:00, 19.71it/s, loss=0] Epoch: 291: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 291: 100% 213/213 [00:10<00:00, 19.69it/s, loss=0] Epoch: 292: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 292: 100% 213/213 [00:10<00:00, 19.70it/s, loss=0] Epoch: 293: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 293: 100% 213/213 [00:10<00:00, 19.65it/s, loss=0] Epoch: 294: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 294: 100% 213/213 [00:10<00:00, 19.69it/s, loss=0] Epoch: 295: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 295: 100% 213/213 [00:10<00:00, 19.68it/s, loss=0] Epoch: 296: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 296: 100% 213/213 [00:10<00:00, 19.70it/s, loss=0] Epoch: 297: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 297: 100% 213/213 [00:10<00:00, 19.75it/s, loss=0] Epoch: 298: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 298: 100% 213/213 [00:10<00:00, 19.85it/s, loss=0] Epoch: 299: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 299: 100% 213/213 [00:10<00:00, 19.77it/s, loss=0] Epoch: 100% 300/300 [54:22<00:00, 10.87s/it] INFO:root:Training is finished ... Iteration: 100% 24/24 [00:10<00:00, 2.25it/s, acc=1] Iteration: 100% 165/165 [16:30<00:00, 6.01s/it, acc=0.889] obj_pp_to_subj_pp: 16.4 cp_recursion: 52.2 pp_recursion: 24.9 subj_to_obj_proper: 83.9 prim_to_obj_proper: 90.2 prim_to_subj_proper: 100.0 LEX: 99.89333333333333 OVERALL: 88.85714285714286
seed 42: OVERALL: 87.36190476190477
seed 43: OVERALL: 90.46666666666667
seed 44: OVERALL: 85.20476190476191
seed 45: OVERALL: 90.67619047619047
seed 66: OVERALL: 89.84761904761905
seed 67: OVERALL: 87.83333333333333
seed 68: OVERALL: 90.32857142857142
seed 69: OVERALL: 89.20476190476191
seed 77: OVERALL: 90.14285714285715
seed 78: OVERALL: 87.5047619047619
seed 79: OVERALL: 90.86666666666666
seed 80: OVERALL: 88.82857142857142
seed 88: OVERALL: 85.04761904761905
seed 89: OVERALL: 89.66190476190476
seed 90: OVERALL: 90.10000000000001
seed 91: OVERALL: 85.41904761904762
seed 99: OVERALL: 85.77142857142857
seed 100: OVERALL: 88.41904761904762
seed 101: OVERALL: 89.4
seed 102: OVERALL: 88.85714285714286
In [ ]:
import numpy as np
wu_et_al_2023_baseline_overall_sem_accuracy_recogs_pos = np.array([87.36190476190477, 90.46666666666667, 85.20476190476191, 90.67619047619047,
89.84761904761905, 87.83333333333333, 90.32857142857142, 89.20476190476191, 90.14285714285715, 87.5047619047619,
90.86666666666666, 88.82857142857142, 85.04761904761905, 89.66190476190476, 90.10000000000001 ,
85.41904761904762, 85.77142857142857, 88.41904761904762, 89.4, 88.85714285714286])
In [ ]:
wu_et_al_2023_baseline_overall_sem_accuracy_recogs_pos.mean()
Out[ ]:
88.54714285714286
In [ ]:
wu_et_al_2023_baseline_overall_sem_accuracy_recogs_pos.std()
Out[ ]:
1.8697974837894837
In [ ]:
import math
wu_et_al_2023_baseline_overall_sem_accuracy_recogs_pos.std()/math.sqrt(len(wu_et_al_2023_baseline_overall_sem_accuracy_recogs_pos))
Out[ ]:
0.4180994277911346
In [ ]:
sem_accuracy_overall_stderr_1p96 = wu_et_al_2023_baseline_overall_sem_accuracy_recogs_pos.std()/math.sqrt(len(wu_et_al_2023_baseline_overall_sem_accuracy_recogs_pos))*1.96
In [ ]:
(wu_et_al_2023_baseline_overall_sem_accuracy_recogs_pos.mean() - sem_accuracy_overall_stderr_1p96, wu_et_al_2023_baseline_overall_sem_accuracy_recogs_pos.mean() + sem_accuracy_overall_stderr_1p96)
Out[ ]:
(87.72766797867223, 89.36661773561349)
(note these are on GPU so re-running with same seeds will not produce same result, but should be from same statistical distribution)
seed 42: obj_pp_to_subj_pp: 14.8
seed 66: obj_pp_to_subj_pp: 19.7
seed 77: obj_pp_to_subj_pp: 31.0
seed 88: obj_pp_to_subj_pp: 13.5
seed 99: obj_pp_to_subj_pp: 17.8
seed 43: obj_pp_to_subj_pp: 20.2
seed 44: obj_pp_to_subj_pp: 15.1
seed 67: obj_pp_to_subj_pp: 17.5
seed 68: obj_pp_to_subj_pp: 16.4
seed 78: obj_pp_to_subj_pp: 20.0
seed 79: obj_pp_to_subj_pp: 23.0
seed 89: obj_pp_to_subj_pp: 20.1
seed 90: obj_pp_to_subj_pp: 12.7
seed 100: obj_pp_to_subj_pp: 15.8
seed 101: obj_pp_to_subj_pp: 12.4
seed 45: obj_pp_to_subj_pp: 31.4
seed 69: obj_pp_to_subj_pp: 35.3
seed 80: obj_pp_to_subj_pp: 21.7
seed 91: obj_pp_to_subj_pp: 18.9
seed 102: obj_pp_to_subj_pp: 16.4
In [ ]:
import numpy as np
wu_et_al_2023_baseline_recogs_pos_subjpp = np.array([14.8, 19.7, 31.0, 13.5, 17.8, 20.2, 15.1, 17.5, 16.4, 20.0, 23.0, 20.1, 12.7, 15.8, 12.4, 31.4, 35.3, 21.7, 18.9, 16.4])
In [ ]:
wu_et_al_2023_baseline_recogs_pos_subjpp.mean()
Out[ ]:
19.684999999999995
In [ ]:
len(wu_et_al_2023_baseline_recogs_pos_subjpp)
Out[ ]:
20
In [ ]:
wu_et_al_2023_baseline_recogs_pos_subjpp.std()
Out[ ]:
6.145345799871639
In [ ]:
import math
stderr_x_1p96 = wu_et_al_2023_baseline_recogs_pos_subjpp.std()/math.sqrt(20)*1.96
In [ ]:
(wu_et_al_2023_baseline_recogs_pos_subjpp.mean() - stderr_x_1p96, wu_et_al_2023_baseline_recogs_pos_subjpp.mean() + stderr_x_1p96)
Out[ ]:
(16.991683453063857, 22.378316546936134)
Wu et al 2023 baseline Encoder-Decoder Transformer with 3 layers, 4 layers (note our claim is these do not do better because like our RASP program it is a flat solution the Transformer has learned, not a tree like, recursive one)¶
see also some early explorations on this at (spread over 3 notebooks so could run in parallel), higher experiment count but not on ReCOGS positional dataset (just plain ReCOGS) so would not have been directly comparable to RASP model runs which were done on ReCOGS_pos so could also report String Exact Match (not just Semantic Exact Match) since RASP model can do it.
https://colab.research.google.com/drive/19_M-KC98vK5_2ZiQj0UR8CkiVwuB-lvO
https://colab.research.google.com/drive/1WvMyX-fngMj5MKm10NP4Jct7H63hXzSQ
https://colab.research.google.com/drive/1X2rRBR8WfBr4zCvDJuuaRsJw5UrIsWOu
These links are not public since not reported in the paper and consistent with findings here.
Wu et al 2023 baseline Encoder-Decoder - 3 layers - not controlling for parameter count¶
ReCOGS commit used 1b6eca8ff4dca5fd2fb284a7d470998af5083beb
In [ ]:
%cd /content/
!rm -rf ReCOGS
!git clone https://github.com/frankaging/ReCOGS.git
%cd ReCOGS
!echo '{\
"architectures": [\
"Bert"\
],\
"model_type": "bert",\
"attention_probs_dropout_prob": 0.1,\
"hidden_act": "gelu",\
"hidden_dropout_prob": 0.1,\
"hidden_size": 300,\
"initializer_range": 0.02,\
"intermediate_size": 512,\
"num_attention_heads": 4,\
"num_hidden_layers": 3,\
"type_vocab_size": 2,\
"vocab_size": 762,\
"pad_token_id": 0,\
"bos_token_id": 1,\
"eos_token_id": 2,\
"unk_token_id": 3,\
"mask_token_id": 4,\
"cls_token_id": 5,\
"sum_token_id": 6,\
"nsp_token_id": 7,\
"position_embedding_type": "absolute",\
"position_embedding_init": "random"\
}' > /content/ReCOGS/model/encoder_config.json
!echo '{\
"architectures": [\
"Bert"\
],\
"model_type": "bert",\
"attention_probs_dropout_prob": 0.1,\
"hidden_act": "gelu",\
"hidden_dropout_prob": 0.1,\
"hidden_size": 300,\
"initializer_range": 0.02,\
"intermediate_size": 512,\
"num_attention_heads": 4,\
"num_hidden_layers": 3,\
"type_vocab_size": 2,\
"vocab_size": 729,\
"pad_token_id": 0,\
"bos_token_id": 1,\
"eos_token_id": 2,\
"unk_token_id": 3,\
"mask_token_id": 4,\
"decoder_start_token_id": 1,\
"position_embedding_type": "absolute",\
"position_embedding_init": "random"\
}' > /content/ReCOGS/model/decoder_config.json
/content Cloning into 'ReCOGS'... remote: Enumerating objects: 436, done. remote: Counting objects: 100% (124/124), done. remote: Compressing objects: 100% (51/51), done. remote: Total 436 (delta 96), reused 92 (delta 73), pack-reused 312 (from 1) Receiving objects: 100% (436/436), 84.71 MiB | 33.61 MiB/s, done. Resolving deltas: 100% (303/303), done. Updating files: 100% (137/137), done. /content/ReCOGS
In [ ]:
!python run_cogs.py --model_name ende_transformer --gpu 1 --train_batch_size 128 --eval_batch_size 128 --lr 0.0001 --data_path ./recogs_positional_index --output_dir ./results_recogs_positional_index --lfs cogs --do_train --do_test --do_gen --max_seq_len 512 --output_json --epochs 300 --seeds "42;66;77;88;99" # paper's seeds, not sure how they chose them
INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 6046701 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.42 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]We strongly recommend passing in an `attention_mask` since your input_ids may be padded. See https://huggingface.co/docs/transformers/troubleshooting#incorrect-output-when-padding-tokens-arent-masked. /content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:16<00:00, 12.78it/s, loss=5.85] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:15<00:00, 14.18it/s, loss=4.53] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:14<00:00, 14.26it/s, loss=3.42] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:14<00:00, 14.27it/s, loss=2.38] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:14<00:00, 14.28it/s, loss=1.82] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:14<00:00, 14.27it/s, loss=1.47] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:14<00:00, 14.29it/s, loss=1.22] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:14<00:00, 14.30it/s, loss=1.04] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.91] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.78] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.67] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.57] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.49] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.41] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.35] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.29] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0.24] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.2] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.16] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.14] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.12] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.11] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.09] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.08] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.08] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.06] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.06] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.05] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.04] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.04] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.04] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.03] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.03] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.03] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.03] Epoch: 35: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 35: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.02] Epoch: 36: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 36: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.02] Epoch: 37: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 37: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.02] Epoch: 38: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 38: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.02] Epoch: 39: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 39: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.02] Epoch: 40: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 40: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.02] Epoch: 41: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 41: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.01] Epoch: 42: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 42: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0.01] Epoch: 43: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 43: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.02] Epoch: 44: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 44: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.02] Epoch: 45: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 45: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.01] Epoch: 46: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 46: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.01] Epoch: 47: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 47: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0.02] Epoch: 48: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 48: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.01] Epoch: 49: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 49: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.01] Epoch: 50: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 50: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.01] Epoch: 51: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 51: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.01] Epoch: 52: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 52: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.01] Epoch: 53: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 53: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.01] Epoch: 54: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 54: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.01] Epoch: 55: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 55: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0.01] Epoch: 56: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 56: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.01] Epoch: 57: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 57: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 58: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 58: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.01] Epoch: 59: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 59: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 60: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 60: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 61: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 61: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.01] Epoch: 62: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 62: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.01] Epoch: 63: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 63: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 64: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 64: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 65: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 65: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.01] Epoch: 66: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 66: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 67: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 67: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.01] Epoch: 68: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 68: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.01] Epoch: 69: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 69: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 70: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 70: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.01] Epoch: 71: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 71: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 72: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 72: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 73: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 73: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 74: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 74: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 75: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 75: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 76: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 76: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 77: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 77: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.01] Epoch: 78: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 78: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 79: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 79: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 80: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 80: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 81: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 81: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 82: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 82: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 83: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 83: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 84: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 84: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 85: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 85: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 86: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 86: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 87: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 87: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 88: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 88: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 89: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 89: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 90: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 90: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 91: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 91: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 92: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 92: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 93: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 93: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 94: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 94: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 95: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 95: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 96: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 96: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 97: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 97: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 98: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 98: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 99: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 99: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 100: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 100: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 101: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 101: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 102: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 102: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 103: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 103: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 104: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 104: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 105: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 105: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 106: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 106: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 107: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 107: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 108: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 108: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 109: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 109: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 110: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 110: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 111: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 111: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 112: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 112: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 113: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 113: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 114: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 114: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 115: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 115: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 116: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 116: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 117: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 117: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 118: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 118: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 119: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 119: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 120: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 120: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 121: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 121: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 122: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 122: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 123: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 123: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 124: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 124: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 125: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 125: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 126: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 126: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 127: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 127: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 128: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 128: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 129: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 129: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 130: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 130: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 131: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 131: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 132: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 132: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 133: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 133: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 134: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 134: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 135: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 135: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 136: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 136: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 137: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 137: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 138: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 138: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 139: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 139: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 140: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 140: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 141: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 141: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 142: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 142: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 143: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 143: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 144: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 144: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 145: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 145: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 146: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 146: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 147: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 147: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 148: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 148: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 149: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 149: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 150: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 150: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 151: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 151: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 152: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 152: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 153: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 153: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 154: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 154: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 155: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 155: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 156: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 156: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 157: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 157: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 158: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 158: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 159: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 159: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 160: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 160: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 161: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 161: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 162: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 162: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 163: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 163: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 164: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 164: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 165: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 165: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 166: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 166: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 167: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 167: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 168: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 168: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 169: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 169: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 170: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 170: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 171: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 171: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 172: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 172: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 173: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 173: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 174: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 174: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 175: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 175: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 176: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 176: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 177: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 177: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 178: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 178: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 179: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 179: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 180: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 180: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 181: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 181: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 182: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 182: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 183: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 183: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 184: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 184: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 185: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 185: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 186: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 186: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 187: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 187: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 188: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 188: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 189: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 189: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 190: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 190: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 191: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 191: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 192: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 192: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 193: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 193: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 194: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 194: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 195: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 195: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 196: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 196: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 197: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 197: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 198: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 198: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 199: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 199: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 200: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 200: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 201: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 201: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 202: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 202: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 203: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 203: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 204: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 204: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 205: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 205: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 206: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 206: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 207: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 207: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 208: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 208: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 209: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 209: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 210: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 210: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 211: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 211: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 212: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 212: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 213: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 213: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 214: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 214: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 215: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 215: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 216: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 216: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 217: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 217: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 218: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 218: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 219: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 219: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 220: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 220: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 221: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 221: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 222: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 222: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 223: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 223: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 224: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 224: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 225: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 225: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 226: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 226: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 227: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 227: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 228: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 228: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 229: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 229: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 230: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 230: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 231: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 231: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 232: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 232: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 233: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 233: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 234: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 234: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 235: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 235: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 236: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 236: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 237: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 237: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 238: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 238: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 239: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 239: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 240: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 240: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 241: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 241: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 242: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 242: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 243: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 243: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 244: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 244: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 245: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 245: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 246: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 246: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 247: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 247: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 248: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 248: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 249: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 249: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 250: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 250: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 251: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 251: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 252: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 252: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 253: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 253: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 254: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 254: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 255: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 255: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 256: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 256: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 257: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 257: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 258: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 258: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 259: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 259: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 260: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 260: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 261: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 261: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 262: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 262: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 263: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 263: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 264: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 264: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 265: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 265: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 266: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 266: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 267: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 267: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 268: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 268: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 269: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 269: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 270: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 270: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 271: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 271: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 272: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 272: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 273: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 273: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 274: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 274: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 275: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 275: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 276: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 276: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 277: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 277: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 278: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 278: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 279: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 279: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 280: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 280: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 281: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 281: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 282: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 282: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 283: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 283: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 284: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 284: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 285: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 285: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 286: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 286: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 287: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 287: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 288: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 288: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 289: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 289: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 290: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 290: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 291: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 291: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 292: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 292: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 293: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 293: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 294: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 294: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 295: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 295: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 296: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 296: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 297: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 297: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 298: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 298: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 299: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 299: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 100% 300/300 [1:14:53<00:00, 14.98s/it] INFO:root:Training is finished ... Iteration: 100% 24/24 [00:13<00:00, 1.72it/s, acc=1] Iteration: 100% 165/165 [29:29<00:00, 10.73s/it, acc=0.885] obj_pp_to_subj_pp: 16.6 cp_recursion: 62.3 pp_recursion: 46.9 subj_to_obj_proper: 80.8 prim_to_obj_proper: 89.9 prim_to_subj_proper: 100.0 LEX: 97.47333333333333 OVERALL: 88.5047619047619 INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 6046701 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.66 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:14<00:00, 14.25it/s, loss=5.9] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:14<00:00, 14.29it/s, loss=4.6] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:14<00:00, 14.33it/s, loss=3.43] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:14<00:00, 14.31it/s, loss=2.39] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:14<00:00, 14.32it/s, loss=1.87] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:14<00:00, 14.30it/s, loss=1.54] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:14<00:00, 14.34it/s, loss=1.28] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:14<00:00, 14.33it/s, loss=1.09] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.96] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.86] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.75] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.66] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.57] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.49] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.42] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.36] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.31] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.26] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.22] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.19] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.15] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.13] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.13] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.11] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.1] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.08] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.08] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.06] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.06] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.05] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.05] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.05] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.05] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0.04] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.03] Epoch: 35: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 35: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.03] Epoch: 36: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 36: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.03] Epoch: 37: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 37: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0.03] Epoch: 38: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 38: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.03] Epoch: 39: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 39: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.03] Epoch: 40: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 40: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.02] Epoch: 41: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 41: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.02] Epoch: 42: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 42: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.02] Epoch: 43: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 43: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.02] Epoch: 44: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 44: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.02] Epoch: 45: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 45: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.02] Epoch: 46: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 46: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.02] Epoch: 47: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 47: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.02] Epoch: 48: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 48: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0.01] Epoch: 49: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 49: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.01] Epoch: 50: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 50: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.01] Epoch: 51: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 51: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.01] Epoch: 52: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 52: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.01] Epoch: 53: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 53: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.01] Epoch: 54: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 54: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.01] Epoch: 55: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 55: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0.01] Epoch: 56: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 56: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.01] Epoch: 57: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 57: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.01] Epoch: 58: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 58: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.01] Epoch: 59: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 59: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0.01] Epoch: 60: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 60: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.01] Epoch: 61: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 61: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.01] Epoch: 62: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 62: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.01] Epoch: 63: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 63: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.01] Epoch: 64: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 64: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.01] Epoch: 65: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 65: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 66: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 66: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.01] Epoch: 67: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 67: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 68: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 68: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 69: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 69: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 70: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 70: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 71: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 71: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.01] Epoch: 72: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 72: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0.01] Epoch: 73: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 73: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.01] Epoch: 74: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 74: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 75: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 75: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.01] Epoch: 76: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 76: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 77: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 77: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 78: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 78: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.01] Epoch: 79: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 79: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.01] Epoch: 80: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 80: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 81: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 81: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 82: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 82: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0.01] Epoch: 83: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 83: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 84: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 84: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 85: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 85: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 86: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 86: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 87: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 87: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 88: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 88: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 89: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 89: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 90: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 90: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 91: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 91: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 92: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 92: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 93: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 93: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 94: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 94: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 95: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 95: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.01] Epoch: 96: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 96: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 97: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 97: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 98: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 98: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 99: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 99: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 100: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 100: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 101: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 101: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 102: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 102: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 103: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 103: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 104: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 104: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 105: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 105: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 106: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 106: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 107: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 107: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 108: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 108: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 109: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 109: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 110: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 110: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 111: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 111: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 112: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 112: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 113: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 113: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 114: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 114: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 115: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 115: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 116: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 116: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 117: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 117: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 118: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 118: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 119: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 119: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 120: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 120: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 121: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 121: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 122: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 122: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 123: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 123: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 124: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 124: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 125: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 125: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 126: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 126: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 127: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 127: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 128: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 128: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 129: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 129: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 130: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 130: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 131: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 131: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 132: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 132: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 133: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 133: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 134: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 134: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 135: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 135: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 136: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 136: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 137: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 137: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 138: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 138: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 139: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 139: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 140: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 140: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 141: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 141: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 142: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 142: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 143: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 143: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 144: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 144: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 145: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 145: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 146: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 146: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 147: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 147: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 148: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 148: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 149: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 149: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 150: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 150: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 151: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 151: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 152: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 152: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 153: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 153: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 154: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 154: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 155: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 155: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 156: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 156: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 157: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 157: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 158: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 158: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 159: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 159: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 160: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 160: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 161: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 161: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 162: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 162: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 163: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 163: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 164: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 164: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 165: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 165: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 166: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 166: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 167: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 167: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 168: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 168: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 169: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 169: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 170: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 170: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 171: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 171: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 172: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 172: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 173: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 173: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 174: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 174: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 175: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 175: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 176: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 176: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 177: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 177: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 178: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 178: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 179: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 179: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 180: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 180: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 181: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 181: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 182: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 182: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 183: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 183: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 184: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 184: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 185: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 185: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 186: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 186: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 187: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 187: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 188: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 188: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 189: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 189: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 190: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 190: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 191: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 191: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 192: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 192: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 193: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 193: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 194: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 194: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 195: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 195: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 196: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 196: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 197: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 197: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 198: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 198: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 199: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 199: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 200: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 200: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 201: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 201: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 202: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 202: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 203: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 203: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 204: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 204: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 205: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 205: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 206: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 206: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 207: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 207: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 208: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 208: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 209: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 209: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 210: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 210: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 211: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 211: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 212: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 212: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 213: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 213: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 214: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 214: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 215: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 215: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 216: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 216: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 217: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 217: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 218: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 218: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 219: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 219: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 220: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 220: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 221: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 221: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 222: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 222: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 223: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 223: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 224: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 224: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 225: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 225: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 226: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 226: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 227: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 227: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 228: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 228: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 229: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 229: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 230: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 230: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 231: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 231: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 232: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 232: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 233: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 233: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 234: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 234: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 235: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 235: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 236: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 236: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 237: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 237: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 238: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 238: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 239: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 239: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 240: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 240: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 241: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 241: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 242: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 242: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 243: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 243: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 244: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 244: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 245: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 245: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 246: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 246: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 247: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 247: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 248: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 248: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 249: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 249: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 250: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 250: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 251: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 251: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 252: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 252: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 253: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 253: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 254: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 254: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 255: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 255: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 256: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 256: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 257: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 257: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 258: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 258: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 259: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 259: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 260: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 260: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 261: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 261: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 262: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 262: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 263: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 263: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 264: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 264: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 265: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 265: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 266: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 266: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 267: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 267: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 268: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 268: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 269: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 269: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 270: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 270: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 271: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 271: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 272: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 272: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 273: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 273: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 274: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 274: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 275: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 275: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 276: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 276: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 277: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 277: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 278: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 278: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 279: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 279: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 280: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 280: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 281: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 281: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 282: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 282: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 283: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 283: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 284: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 284: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 285: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 285: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 286: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 286: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 287: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 287: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 288: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 288: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 289: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 289: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 290: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 290: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 291: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 291: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 292: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 292: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 293: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 293: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 294: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 294: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 295: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 295: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 296: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 296: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 297: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 297: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 298: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 298: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 299: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 299: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 100% 300/300 [1:14:55<00:00, 14.99s/it] INFO:root:Training is finished ... Iteration: 100% 24/24 [00:14<00:00, 1.70it/s, acc=1] Iteration: 100% 165/165 [25:46<00:00, 9.37s/it, acc=0.889] obj_pp_to_subj_pp: 13.0 cp_recursion: 55.7 pp_recursion: 12.4 subj_to_obj_proper: 90.6 prim_to_obj_proper: 95.7 prim_to_subj_proper: 99.9 LEX: 99.92666666666666 OVERALL: 88.86666666666667 INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 6046701 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.77 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:15<00:00, 14.18it/s, loss=5.79] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:14<00:00, 14.28it/s, loss=4.55] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:14<00:00, 14.26it/s, loss=3.53] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:14<00:00, 14.27it/s, loss=2.48] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:14<00:00, 14.27it/s, loss=1.93] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:14<00:00, 14.29it/s, loss=1.61] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:14<00:00, 14.25it/s, loss=1.34] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:14<00:00, 14.30it/s, loss=1.16] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:14<00:00, 14.28it/s, loss=1.02] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.91] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0.81] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.72] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0.63] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.55] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.48] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0.41] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.35] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.3] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.26] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0.22] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.2] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.17] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.15] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0.13] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.13] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.11] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.1] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0.1] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.09] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0.08] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.08] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0.07] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.07] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.06] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.06] Epoch: 35: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 35: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0.05] Epoch: 36: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 36: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.05] Epoch: 37: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 37: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.05] Epoch: 38: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 38: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.04] Epoch: 39: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 39: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0.04] Epoch: 40: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 40: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.04] Epoch: 41: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 41: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0.04] Epoch: 42: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 42: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.03] Epoch: 43: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 43: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0.03] Epoch: 44: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 44: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.03] Epoch: 45: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 45: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0.03] Epoch: 46: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 46: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.03] Epoch: 47: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 47: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0.02] Epoch: 48: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 48: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0.02] Epoch: 49: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 49: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0.02] Epoch: 50: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 50: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0.02] Epoch: 51: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 51: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.02] Epoch: 52: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 52: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.02] Epoch: 53: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 53: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.02] Epoch: 54: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 54: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.02] Epoch: 55: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 55: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0.01] Epoch: 56: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 56: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.02] Epoch: 57: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 57: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.02] Epoch: 58: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 58: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.01] Epoch: 59: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 59: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0.02] Epoch: 60: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 60: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.01] Epoch: 61: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 61: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.02] Epoch: 62: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 62: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.01] Epoch: 63: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 63: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.02] Epoch: 64: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 64: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0.01] Epoch: 65: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 65: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.01] Epoch: 66: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 66: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.01] Epoch: 67: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 67: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.01] Epoch: 68: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 68: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0.01] Epoch: 69: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 69: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.01] Epoch: 70: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 70: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.01] Epoch: 71: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 71: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.01] Epoch: 72: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 72: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0.01] Epoch: 73: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 73: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.01] Epoch: 74: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 74: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.01] Epoch: 75: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 75: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0.01] Epoch: 76: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 76: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.01] Epoch: 77: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 77: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.01] Epoch: 78: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 78: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0.01] Epoch: 79: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 79: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0.01] Epoch: 80: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 80: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0.01] Epoch: 81: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 81: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0.01] Epoch: 82: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 82: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0.01] Epoch: 83: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 83: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 84: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 84: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0.01] Epoch: 85: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 85: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0.01] Epoch: 86: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 86: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 87: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 87: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.01] Epoch: 88: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 88: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0.01] Epoch: 89: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 89: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.01] Epoch: 90: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 90: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.01] Epoch: 91: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 91: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.01] Epoch: 92: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 92: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0.01] Epoch: 93: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 93: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 94: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 94: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 95: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 95: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 96: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 96: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 97: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 97: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 98: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 98: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 99: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 99: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 100: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 100: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 101: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 101: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 102: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 102: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 103: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 103: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.01] Epoch: 104: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 104: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0.01] Epoch: 105: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 105: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 106: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 106: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 107: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 107: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 108: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 108: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 109: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 109: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 110: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 110: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 111: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 111: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 112: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 112: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 113: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 113: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 114: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 114: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 115: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 115: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 116: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 116: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 117: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 117: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 118: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 118: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 119: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 119: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 120: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 120: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 121: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 121: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 122: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 122: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 123: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 123: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 124: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 124: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 125: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 125: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 126: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 126: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.01] Epoch: 127: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 127: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 128: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 128: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 129: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 129: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 130: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 130: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 131: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 131: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 132: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 132: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 133: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 133: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 134: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 134: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 135: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 135: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 136: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 136: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 137: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 137: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 138: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 138: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 139: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 139: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 140: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 140: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 141: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 141: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 142: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 142: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 143: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 143: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 144: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 144: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 145: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 145: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 146: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 146: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 147: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 147: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 148: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 148: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 149: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 149: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 150: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 150: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 151: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 151: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 152: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 152: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 153: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 153: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 154: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 154: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 155: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 155: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 156: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 156: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 157: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 157: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 158: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 158: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 159: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 159: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 160: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 160: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 161: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 161: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 162: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 162: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 163: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 163: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 164: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 164: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 165: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 165: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 166: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 166: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 167: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 167: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 168: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 168: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 169: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 169: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 170: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 170: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 171: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 171: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 172: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 172: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 173: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 173: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 174: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 174: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 175: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 175: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 176: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 176: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 177: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 177: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 178: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 178: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 179: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 179: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 180: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 180: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 181: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 181: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 182: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 182: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 183: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 183: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 184: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 184: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 185: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 185: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 186: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 186: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 187: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 187: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 188: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 188: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 189: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 189: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 190: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 190: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 191: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 191: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 192: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 192: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 193: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 193: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 194: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 194: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 195: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 195: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 196: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 196: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 197: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 197: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 198: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 198: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 199: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 199: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 200: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 200: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 201: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 201: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 202: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 202: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 203: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 203: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 204: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 204: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 205: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 205: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 206: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 206: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 207: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 207: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 208: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 208: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 209: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 209: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 210: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 210: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 211: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 211: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 212: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 212: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 213: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 213: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 214: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 214: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 215: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 215: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 216: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 216: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 217: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 217: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 218: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 218: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 219: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 219: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 220: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 220: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 221: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 221: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 222: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 222: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 223: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 223: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 224: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 224: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 225: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 225: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 226: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 226: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 227: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 227: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 228: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 228: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 229: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 229: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 230: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 230: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 231: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 231: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 232: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 232: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 233: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 233: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 234: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 234: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 235: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 235: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 236: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 236: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 237: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 237: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 238: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 238: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 239: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 239: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 240: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 240: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 241: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 241: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 242: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 242: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 243: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 243: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 244: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 244: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 245: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 245: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 246: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 246: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 247: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 247: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 248: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 248: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 249: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 249: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 250: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 250: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 251: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 251: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 252: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 252: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 253: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 253: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 254: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 254: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 255: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 255: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 256: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 256: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 257: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 257: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 258: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 258: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 259: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 259: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 260: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 260: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 261: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 261: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 262: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 262: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 263: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 263: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 264: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 264: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 265: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 265: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 266: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 266: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 267: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 267: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 268: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 268: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 269: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 269: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 270: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 270: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 271: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 271: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 272: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 272: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 273: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 273: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 274: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 274: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 275: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 275: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 276: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 276: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 277: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 277: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 278: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 278: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 279: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 279: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 280: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 280: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 281: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 281: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 282: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 282: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 283: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 283: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 284: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 284: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 285: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 285: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 286: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 286: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 287: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 287: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 288: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 288: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 289: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 289: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 290: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 290: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 291: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 291: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 292: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 292: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 293: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 293: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 294: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 294: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 295: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 295: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 296: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 296: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 297: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 297: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 298: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 298: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 299: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 299: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 100% 300/300 [1:15:08<00:00, 15.03s/it] INFO:root:Training is finished ... Iteration: 100% 24/24 [00:14<00:00, 1.69it/s, acc=1] Iteration: 100% 165/165 [24:39<00:00, 8.97s/it, acc=0.9] obj_pp_to_subj_pp: 17.0 cp_recursion: 58.6 pp_recursion: 48.4 subj_to_obj_proper: 93.8 prim_to_obj_proper: 92.5 prim_to_subj_proper: 100.0 LEX: 98.71333333333332 OVERALL: 90.04761904761904 INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 6046701 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.88 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:14<00:00, 14.29it/s, loss=5.79] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:14<00:00, 14.35it/s, loss=4.51] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:14<00:00, 14.32it/s, loss=3.34] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:14<00:00, 14.31it/s, loss=2.28] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:14<00:00, 14.33it/s, loss=1.75] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:14<00:00, 14.34it/s, loss=1.4] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:14<00:00, 14.34it/s, loss=1.12] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.94] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.81] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:14<00:00, 14.37it/s, loss=0.69] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0.58] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.49] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.41] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0.34] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.27] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0.22] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0.17] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0.14] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:14<00:00, 14.37it/s, loss=0.1] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.08] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.08] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0.05] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:14<00:00, 14.37it/s, loss=0.05] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:14<00:00, 14.39it/s, loss=0.04] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0.03] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0.03] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:14<00:00, 14.38it/s, loss=0.02] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0.02] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.02] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:14<00:00, 14.38it/s, loss=0.02] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:14<00:00, 14.38it/s, loss=0.01] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:14<00:00, 14.37it/s, loss=0.01] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.01] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.01] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.01] Epoch: 35: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 35: 100% 213/213 [00:14<00:00, 14.38it/s, loss=0.01] Epoch: 36: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 36: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.01] Epoch: 37: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 37: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 38: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 38: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 39: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 39: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 40: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 40: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 41: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 41: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 42: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 42: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 43: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 43: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 44: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 44: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 45: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 45: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 46: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 46: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 47: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 47: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 48: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 48: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 49: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 49: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 50: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 50: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 51: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 51: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 52: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 52: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 53: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 53: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 54: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 54: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 55: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 55: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 56: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 56: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 57: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 57: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 58: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 58: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 59: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 59: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 60: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 60: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 61: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 61: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 62: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 62: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 63: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 63: 100% 213/213 [00:14<00:00, 14.38it/s, loss=0] Epoch: 64: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 64: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0.01] Epoch: 65: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 65: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 66: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 66: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 67: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 67: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 68: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 68: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 69: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 69: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 70: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 70: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 71: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 71: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 72: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 72: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 73: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 73: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 74: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 74: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 75: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 75: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 76: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 76: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 77: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 77: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 78: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 78: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 79: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 79: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 80: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 80: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 81: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 81: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 82: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 82: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 83: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 83: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 84: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 84: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 85: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 85: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 86: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 86: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 87: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 87: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 88: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 88: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 89: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 89: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 90: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 90: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 91: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 91: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 92: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 92: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 93: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 93: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 94: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 94: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 95: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 95: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 96: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 96: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 97: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 97: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 98: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 98: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 99: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 99: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 100: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 100: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 101: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 101: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 102: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 102: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 103: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 103: 100% 213/213 [00:14<00:00, 14.37it/s, loss=0] Epoch: 104: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 104: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 105: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 105: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 106: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 106: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 107: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 107: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 108: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 108: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 109: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 109: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 110: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 110: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 111: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 111: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 112: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 112: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 113: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 113: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 114: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 114: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 115: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 115: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 116: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 116: 100% 213/213 [00:14<00:00, 14.37it/s, loss=0] Epoch: 117: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 117: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 118: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 118: 100% 213/213 [00:14<00:00, 14.37it/s, loss=0] Epoch: 119: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 119: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 120: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 120: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 121: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 121: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 122: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 122: 100% 213/213 [00:14<00:00, 14.37it/s, loss=0] Epoch: 123: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 123: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 124: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 124: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 125: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 125: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 126: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 126: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0.01] Epoch: 127: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 127: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 128: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 128: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 129: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 129: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 130: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 130: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 131: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 131: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 132: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 132: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 133: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 133: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 134: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 134: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 135: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 135: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 136: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 136: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 137: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 137: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 138: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 138: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 139: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 139: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 140: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 140: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 141: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 141: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 142: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 142: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 143: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 143: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 144: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 144: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 145: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 145: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 146: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 146: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 147: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 147: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 148: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 148: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 149: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 149: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 150: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 150: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 151: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 151: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 152: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 152: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 153: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 153: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 154: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 154: 100% 213/213 [00:14<00:00, 14.37it/s, loss=0] Epoch: 155: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 155: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 156: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 156: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 157: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 157: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 158: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 158: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 159: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 159: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 160: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 160: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 161: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 161: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 162: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 162: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 163: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 163: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 164: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 164: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 165: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 165: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 166: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 166: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 167: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 167: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 168: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 168: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 169: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 169: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 170: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 170: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 171: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 171: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 172: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 172: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 173: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 173: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 174: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 174: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 175: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 175: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 176: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 176: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 177: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 177: 100% 213/213 [00:14<00:00, 14.37it/s, loss=0] Epoch: 178: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 178: 100% 213/213 [00:14<00:00, 14.37it/s, loss=0] Epoch: 179: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 179: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 180: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 180: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 181: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 181: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 182: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 182: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 183: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 183: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 184: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 184: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 185: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 185: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 186: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 186: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 187: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 187: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 188: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 188: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 189: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 189: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 190: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 190: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 191: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 191: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 192: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 192: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 193: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 193: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 194: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 194: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 195: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 195: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 196: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 196: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 197: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 197: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 198: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 198: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 199: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 199: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 200: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 200: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 201: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 201: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 202: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 202: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 203: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 203: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 204: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 204: 100% 213/213 [00:14<00:00, 14.38it/s, loss=0] Epoch: 205: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 205: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 206: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 206: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 207: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 207: 100% 213/213 [00:14<00:00, 14.38it/s, loss=0] Epoch: 208: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 208: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 209: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 209: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 210: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 210: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 211: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 211: 100% 213/213 [00:14<00:00, 14.38it/s, loss=0] Epoch: 212: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 212: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 213: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 213: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 214: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 214: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 215: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 215: 100% 213/213 [00:14<00:00, 14.38it/s, loss=0] Epoch: 216: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 216: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 217: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 217: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 218: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 218: 100% 213/213 [00:14<00:00, 14.38it/s, loss=0] Epoch: 219: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 219: 100% 213/213 [00:14<00:00, 14.37it/s, loss=0] Epoch: 220: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 220: 100% 213/213 [00:14<00:00, 14.38it/s, loss=0] Epoch: 221: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 221: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 222: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 222: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 223: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 223: 100% 213/213 [00:14<00:00, 14.38it/s, loss=0] Epoch: 224: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 224: 100% 213/213 [00:14<00:00, 14.38it/s, loss=0] Epoch: 225: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 225: 100% 213/213 [00:14<00:00, 14.37it/s, loss=0] Epoch: 226: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 226: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 227: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 227: 100% 213/213 [00:14<00:00, 14.37it/s, loss=0] Epoch: 228: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 228: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 229: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 229: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 230: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 230: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 231: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 231: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 232: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 232: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 233: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 233: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 234: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 234: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 235: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 235: 100% 213/213 [00:14<00:00, 14.38it/s, loss=0] Epoch: 236: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 236: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 237: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 237: 100% 213/213 [00:14<00:00, 14.37it/s, loss=0] Epoch: 238: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 238: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 239: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 239: 100% 213/213 [00:14<00:00, 14.38it/s, loss=0] Epoch: 240: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 240: 100% 213/213 [00:14<00:00, 14.38it/s, loss=0] Epoch: 241: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 241: 100% 213/213 [00:14<00:00, 14.38it/s, loss=0] Epoch: 242: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 242: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 243: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 243: 100% 213/213 [00:14<00:00, 14.39it/s, loss=0] Epoch: 244: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 244: 100% 213/213 [00:14<00:00, 14.38it/s, loss=0] Epoch: 245: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 245: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 246: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 246: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 247: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 247: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 248: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 248: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 249: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 249: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 250: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 250: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 251: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 251: 100% 213/213 [00:14<00:00, 14.38it/s, loss=0] Epoch: 252: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 252: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 253: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 253: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 254: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 254: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 255: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 255: 100% 213/213 [00:14<00:00, 14.38it/s, loss=0] Epoch: 256: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 256: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 257: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 257: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 258: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 258: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 259: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 259: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 260: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 260: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 261: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 261: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 262: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 262: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 263: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 263: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 264: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 264: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 265: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 265: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 266: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 266: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 267: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 267: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 268: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 268: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 269: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 269: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 270: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 270: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 271: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 271: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 272: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 272: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 273: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 273: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 274: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 274: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 275: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 275: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 276: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 276: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 277: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 277: 100% 213/213 [00:14<00:00, 14.37it/s, loss=0] Epoch: 278: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 278: 100% 213/213 [00:14<00:00, 14.38it/s, loss=0] Epoch: 279: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 279: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 280: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 280: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 281: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 281: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 282: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 282: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 283: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 283: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 284: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 284: 100% 213/213 [00:14<00:00, 14.38it/s, loss=0] Epoch: 285: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 285: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 286: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 286: 100% 213/213 [00:14<00:00, 14.37it/s, loss=0] Epoch: 287: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 287: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 288: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 288: 100% 213/213 [00:14<00:00, 14.38it/s, loss=0] Epoch: 289: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 289: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 290: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 290: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 291: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 291: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 292: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 292: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 293: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 293: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 294: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 294: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 295: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 295: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 296: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 296: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 297: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 297: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 298: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 298: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 299: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 299: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 100% 300/300 [1:14:39<00:00, 14.93s/it] INFO:root:Training is finished ... Iteration: 100% 24/24 [00:14<00:00, 1.68it/s, acc=1] Iteration: 100% 165/165 [21:26<00:00, 7.80s/it, acc=0.909] obj_pp_to_subj_pp: 16.2 cp_recursion: 57.4 pp_recursion: 64.5 subj_to_obj_proper: 80.4 prim_to_obj_proper: 91.2 prim_to_subj_proper: 100.0 LEX: 99.95333333333333 OVERALL: 90.9047619047619 INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 6046701 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.99 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:15<00:00, 14.13it/s, loss=5.76] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:15<00:00, 14.19it/s, loss=4.52] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:15<00:00, 14.18it/s, loss=3.52] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:15<00:00, 14.16it/s, loss=2.44] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:15<00:00, 14.19it/s, loss=1.87] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:14<00:00, 14.22it/s, loss=1.49] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:15<00:00, 14.19it/s, loss=1.22] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:15<00:00, 14.16it/s, loss=1.04] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0.91] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0.8] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0.71] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0.62] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0.54] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0.46] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0.38] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0.32] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0.27] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0.23] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0.17] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:15<00:00, 14.17it/s, loss=0.16] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0.12] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0.1] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0.08] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:15<00:00, 14.10it/s, loss=0.07] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:15<00:00, 14.13it/s, loss=0.06] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:15<00:00, 14.13it/s, loss=0.05] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:15<00:00, 14.14it/s, loss=0.04] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:15<00:00, 14.10it/s, loss=0.03] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:15<00:00, 14.17it/s, loss=0.03] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0.02] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0.02] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:15<00:00, 14.13it/s, loss=0.02] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:15<00:00, 14.17it/s, loss=0.02] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0.01] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0.01] Epoch: 35: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 35: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0.01] Epoch: 36: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 36: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0.01] Epoch: 37: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 37: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0.01] Epoch: 38: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 38: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0.01] Epoch: 39: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 39: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0.01] Epoch: 40: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 40: 100% 213/213 [00:15<00:00, 14.17it/s, loss=0.01] Epoch: 41: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 41: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0.01] Epoch: 42: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 42: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0.01] Epoch: 43: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 43: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 44: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 44: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0.01] Epoch: 45: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 45: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0.01] Epoch: 46: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 46: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0.01] Epoch: 47: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 47: 100% 213/213 [00:15<00:00, 14.17it/s, loss=0.01] Epoch: 48: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 48: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 49: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 49: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 50: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 50: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 51: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 51: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0.01] Epoch: 52: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 52: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0] Epoch: 53: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 53: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 54: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 54: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 55: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 55: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 56: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 56: 100% 213/213 [00:15<00:00, 14.15it/s, loss=0] Epoch: 57: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 57: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0.01] Epoch: 58: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 58: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 59: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 59: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 60: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 60: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 61: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 61: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 62: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 62: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 63: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 63: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0] Epoch: 64: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 64: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0] Epoch: 65: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 65: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0] Epoch: 66: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 66: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 67: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 67: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 68: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 68: 100% 213/213 [00:15<00:00, 14.14it/s, loss=0] Epoch: 69: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 69: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 70: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 70: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 71: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 71: 100% 213/213 [00:15<00:00, 14.17it/s, loss=0] Epoch: 72: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 72: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 73: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 73: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0] Epoch: 74: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 74: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 75: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 75: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0.01] Epoch: 76: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 76: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 77: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 77: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0] Epoch: 78: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 78: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 79: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 79: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 80: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 80: 100% 213/213 [00:15<00:00, 14.17it/s, loss=0] Epoch: 81: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 81: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 82: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 82: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 83: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 83: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 84: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 84: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 85: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 85: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 86: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 86: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 87: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 87: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 88: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 88: 100% 213/213 [00:15<00:00, 14.15it/s, loss=0.01] Epoch: 89: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 89: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 90: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 90: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 91: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 91: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0] Epoch: 92: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 92: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0] Epoch: 93: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 93: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 94: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 94: 100% 213/213 [00:15<00:00, 14.17it/s, loss=0] Epoch: 95: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 95: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 96: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 96: 100% 213/213 [00:15<00:00, 14.17it/s, loss=0] Epoch: 97: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 97: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 98: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 98: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 99: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 99: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 100: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 100: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 101: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 101: 100% 213/213 [00:15<00:00, 14.17it/s, loss=0] Epoch: 102: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 102: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 103: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 103: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 104: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 104: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 105: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 105: 100% 213/213 [00:15<00:00, 14.15it/s, loss=0] Epoch: 106: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 106: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 107: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 107: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0] Epoch: 108: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 108: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 109: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 109: 100% 213/213 [00:15<00:00, 14.15it/s, loss=0] Epoch: 110: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 110: 100% 213/213 [00:15<00:00, 14.17it/s, loss=0] Epoch: 111: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 111: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 112: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 112: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 113: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 113: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0] Epoch: 114: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 114: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 115: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 115: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 116: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 116: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 117: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 117: 100% 213/213 [00:15<00:00, 14.14it/s, loss=0] Epoch: 118: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 118: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 119: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 119: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 120: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 120: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0] Epoch: 121: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 121: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0] Epoch: 122: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 122: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 123: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 123: 100% 213/213 [00:15<00:00, 14.17it/s, loss=0] Epoch: 124: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 124: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0] Epoch: 125: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 125: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 126: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 126: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 127: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 127: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 128: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 128: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 129: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 129: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0] Epoch: 130: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 130: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 131: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 131: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 132: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 132: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 133: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 133: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0] Epoch: 134: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 134: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 135: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 135: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 136: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 136: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 137: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 137: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0] Epoch: 138: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 138: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 139: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 139: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 140: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 140: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 141: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 141: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0] Epoch: 142: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 142: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 143: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 143: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 144: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 144: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 145: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 145: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 146: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 146: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 147: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 147: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 148: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 148: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 149: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 149: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0] Epoch: 150: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 150: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 151: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 151: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 152: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 152: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 153: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 153: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0] Epoch: 154: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 154: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 155: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 155: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0] Epoch: 156: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 156: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 157: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 157: 100% 213/213 [00:15<00:00, 14.15it/s, loss=0] Epoch: 158: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 158: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0] Epoch: 159: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 159: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 160: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 160: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 161: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 161: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 162: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 162: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0] Epoch: 163: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 163: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 164: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 164: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 165: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 165: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 166: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 166: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 167: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 167: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 168: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 168: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 169: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 169: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 170: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 170: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0] Epoch: 171: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 171: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 172: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 172: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 173: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 173: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 174: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 174: 100% 213/213 [00:15<00:00, 14.15it/s, loss=0] Epoch: 175: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 175: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 176: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 176: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0] Epoch: 177: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 177: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 178: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 178: 100% 213/213 [00:15<00:00, 14.17it/s, loss=0] Epoch: 179: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 179: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 180: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 180: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0] Epoch: 181: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 181: 100% 213/213 [00:15<00:00, 14.17it/s, loss=0] Epoch: 182: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 182: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0] Epoch: 183: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 183: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 184: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 184: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 185: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 185: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 186: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 186: 100% 213/213 [00:15<00:00, 14.17it/s, loss=0] Epoch: 187: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 187: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 188: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 188: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0] Epoch: 189: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 189: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0] Epoch: 190: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 190: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 191: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 191: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 192: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 192: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0] Epoch: 193: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 193: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0] Epoch: 194: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 194: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 195: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 195: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0] Epoch: 196: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 196: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 197: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 197: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 198: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 198: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0] Epoch: 199: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 199: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 200: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 200: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0] Epoch: 201: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 201: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0] Epoch: 202: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 202: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 203: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 203: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 204: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 204: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 205: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 205: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0] Epoch: 206: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 206: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 207: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 207: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 208: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 208: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 209: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 209: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0] Epoch: 210: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 210: 100% 213/213 [00:15<00:00, 14.14it/s, loss=0] Epoch: 211: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 211: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 212: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 212: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 213: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 213: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 214: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 214: 100% 213/213 [00:15<00:00, 14.17it/s, loss=0] Epoch: 215: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 215: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 216: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 216: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0] Epoch: 217: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 217: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 218: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 218: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 219: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 219: 100% 213/213 [00:15<00:00, 14.15it/s, loss=0] Epoch: 220: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 220: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 221: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 221: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 222: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 222: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 223: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 223: 100% 213/213 [00:15<00:00, 14.15it/s, loss=0] Epoch: 224: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 224: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 225: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 225: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 226: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 226: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 227: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 227: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 228: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 228: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 229: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 229: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 230: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 230: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 231: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 231: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0] Epoch: 232: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 232: 100% 213/213 [00:15<00:00, 14.17it/s, loss=0] Epoch: 233: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 233: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 234: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 234: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 235: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 235: 100% 213/213 [00:15<00:00, 14.17it/s, loss=0] Epoch: 236: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 236: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 237: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 237: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 238: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 238: 100% 213/213 [00:15<00:00, 14.17it/s, loss=0] Epoch: 239: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 239: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 240: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 240: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 241: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 241: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 242: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 242: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 243: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 243: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 244: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 244: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 245: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 245: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0] Epoch: 246: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 246: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 247: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 247: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0] Epoch: 248: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 248: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 249: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 249: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 250: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 250: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 251: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 251: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 252: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 252: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 253: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 253: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 254: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 254: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 255: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 255: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 256: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 256: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 257: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 257: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 258: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 258: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 259: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 259: 100% 213/213 [00:15<00:00, 14.17it/s, loss=0] Epoch: 260: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 260: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 261: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 261: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 262: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 262: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 263: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 263: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0] Epoch: 264: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 264: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 265: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 265: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0] Epoch: 266: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 266: 100% 213/213 [00:15<00:00, 14.17it/s, loss=0] Epoch: 267: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 267: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0] Epoch: 268: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 268: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 269: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 269: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 270: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 270: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 271: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 271: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 272: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 272: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 273: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 273: 100% 213/213 [00:15<00:00, 14.17it/s, loss=0] Epoch: 274: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 274: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 275: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 275: 100% 213/213 [00:15<00:00, 14.15it/s, loss=0] Epoch: 276: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 276: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 277: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 277: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 278: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 278: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 279: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 279: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 280: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 280: 100% 213/213 [00:15<00:00, 14.15it/s, loss=0] Epoch: 281: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 281: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 282: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 282: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 283: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 283: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 284: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 284: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0] Epoch: 285: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 285: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 286: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 286: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 287: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 287: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 288: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 288: 100% 213/213 [00:15<00:00, 14.17it/s, loss=0] Epoch: 289: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 289: 100% 213/213 [00:15<00:00, 14.11it/s, loss=0] Epoch: 290: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 290: 100% 213/213 [00:15<00:00, 14.12it/s, loss=0] Epoch: 291: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 291: 100% 213/213 [00:15<00:00, 14.15it/s, loss=0] Epoch: 292: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 292: 100% 213/213 [00:15<00:00, 14.15it/s, loss=0] Epoch: 293: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 293: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0] Epoch: 294: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 294: 100% 213/213 [00:15<00:00, 14.15it/s, loss=0] Epoch: 295: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 295: 100% 213/213 [00:15<00:00, 14.15it/s, loss=0] Epoch: 296: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 296: 100% 213/213 [00:15<00:00, 14.14it/s, loss=0] Epoch: 297: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 297: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0] Epoch: 298: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 298: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 299: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 299: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 100% 300/300 [1:15:31<00:00, 15.10s/it] INFO:root:Training is finished ... Iteration: 100% 24/24 [00:14<00:00, 1.69it/s, acc=1] Iteration: 100% 165/165 [32:06<00:00, 11.68s/it, acc=0.876] obj_pp_to_subj_pp: 13.7 cp_recursion: 58.5 pp_recursion: 49.3 subj_to_obj_proper: 95.9 prim_to_obj_proper: 64.7 prim_to_subj_proper: 99.5 LEX: 97.25333333333333 OVERALL: 87.63809523809523
In [ ]:
!python run_cogs.py --model_name ende_transformer --gpu 1 --train_batch_size 128 --eval_batch_size 128 --lr 0.0001 --data_path ./recogs_positional_index --output_dir ./results_recogs_positional_index --lfs cogs --do_train --do_test --do_gen --max_seq_len 512 --output_json --epochs 300 --seeds "43;67;78;89;100" # paper's seeds, not sure how they chose them
INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 6046701 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.43 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]We strongly recommend passing in an `attention_mask` since your input_ids may be padded. See https://huggingface.co/docs/transformers/troubleshooting#incorrect-output-when-padding-tokens-arent-masked. /content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:15<00:00, 13.42it/s, loss=5.87] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:14<00:00, 14.20it/s, loss=4.58] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:14<00:00, 14.26it/s, loss=3.49] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:14<00:00, 14.26it/s, loss=2.4] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:14<00:00, 14.24it/s, loss=1.83] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:14<00:00, 14.22it/s, loss=1.47] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:14<00:00, 14.25it/s, loss=1.18] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:14<00:00, 14.24it/s, loss=1.01] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0.89] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0.78] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0.69] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0.6] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0.51] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0.43] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0.36] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0.3] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0.24] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0.19] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0.14] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0.13] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0.1] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0.08] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0.07] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0.05] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0.04] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0.04] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0.03] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0.03] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0.02] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.03] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0.02] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0.02] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0.01] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0.01] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0.01] Epoch: 35: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 35: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.01] Epoch: 36: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 36: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.01] Epoch: 37: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 37: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0.01] Epoch: 38: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 38: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0.01] Epoch: 39: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 39: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.01] Epoch: 40: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 40: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.01] Epoch: 41: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 41: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0.01] Epoch: 42: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 42: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0.01] Epoch: 43: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 43: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 44: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 44: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 45: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 45: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 46: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 46: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 47: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 47: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 48: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 48: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 49: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 49: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0.01] Epoch: 50: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 50: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0.01] Epoch: 51: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 51: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 52: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 52: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 53: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 53: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 54: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 54: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0.01] Epoch: 55: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 55: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 56: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 56: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 57: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 57: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 58: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 58: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 59: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 59: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 60: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 60: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 61: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 61: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 62: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 62: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 63: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 63: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0.01] Epoch: 64: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 64: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 65: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 65: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 66: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 66: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 67: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 67: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0.01] Epoch: 68: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 68: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 69: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 69: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 70: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 70: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 71: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 71: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 72: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 72: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 73: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 73: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0.01] Epoch: 74: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 74: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 75: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 75: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0] Epoch: 76: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 76: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 77: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 77: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 78: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 78: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 79: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 79: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0] Epoch: 80: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 80: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 81: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 81: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 82: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 82: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 83: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 83: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 84: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 84: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 85: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 85: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 86: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 86: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 87: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 87: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 88: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 88: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 89: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 89: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 90: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 90: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 91: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 91: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 92: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 92: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 93: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 93: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 94: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 94: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 95: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 95: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 96: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 96: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 97: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 97: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 98: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 98: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 99: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 99: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 100: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 100: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 101: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 101: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 102: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 102: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 103: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 103: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 104: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 104: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 105: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 105: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 106: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 106: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 107: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 107: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 108: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 108: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 109: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 109: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 110: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 110: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 111: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 111: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 112: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 112: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 113: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 113: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 114: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 114: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 115: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 115: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 116: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 116: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 117: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 117: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 118: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 118: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 119: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 119: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 120: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 120: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 121: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 121: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 122: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 122: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 123: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 123: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 124: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 124: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 125: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 125: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 126: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 126: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 127: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 127: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 128: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 128: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 129: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 129: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 130: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 130: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 131: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 131: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 132: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 132: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 133: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 133: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 134: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 134: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 135: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 135: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 136: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 136: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 137: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 137: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 138: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 138: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 139: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 139: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 140: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 140: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 141: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 141: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 142: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 142: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 143: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 143: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 144: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 144: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 145: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 145: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 146: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 146: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 147: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 147: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 148: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 148: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 149: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 149: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 150: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 150: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 151: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 151: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 152: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 152: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 153: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 153: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 154: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 154: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 155: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 155: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 156: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 156: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 157: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 157: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 158: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 158: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 159: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 159: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 160: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 160: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 161: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 161: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 162: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 162: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 163: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 163: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 164: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 164: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 165: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 165: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 166: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 166: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 167: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 167: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 168: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 168: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 169: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 169: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 170: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 170: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 171: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 171: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 172: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 172: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 173: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 173: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 174: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 174: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 175: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 175: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 176: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 176: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 177: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 177: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 178: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 178: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 179: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 179: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 180: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 180: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 181: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 181: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 182: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 182: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 183: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 183: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 184: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 184: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 185: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 185: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 186: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 186: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 187: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 187: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 188: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 188: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 189: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 189: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 190: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 190: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 191: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 191: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 192: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 192: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 193: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 193: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 194: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 194: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 195: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 195: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 196: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 196: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 197: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 197: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 198: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 198: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 199: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 199: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 200: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 200: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 201: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 201: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 202: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 202: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 203: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 203: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 204: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 204: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 205: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 205: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 206: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 206: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 207: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 207: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 208: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 208: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 209: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 209: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 210: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 210: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 211: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 211: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 212: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 212: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 213: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 213: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 214: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 214: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 215: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 215: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 216: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 216: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 217: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 217: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 218: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 218: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 219: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 219: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 220: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 220: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 221: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 221: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 222: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 222: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 223: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 223: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 224: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 224: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 225: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 225: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 226: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 226: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0] Epoch: 227: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 227: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 228: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 228: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 229: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 229: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 230: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 230: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0] Epoch: 231: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 231: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 232: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 232: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 233: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 233: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 234: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 234: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 235: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 235: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 236: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 236: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 237: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 237: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 238: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 238: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 239: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 239: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 240: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 240: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 241: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 241: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 242: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 242: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 243: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 243: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 244: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 244: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 245: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 245: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 246: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 246: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 247: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 247: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 248: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 248: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 249: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 249: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 250: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 250: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 251: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 251: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 252: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 252: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 253: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 253: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 254: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 254: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 255: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 255: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 256: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 256: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 257: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 257: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 258: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 258: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 259: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 259: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 260: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 260: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 261: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 261: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 262: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 262: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 263: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 263: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 264: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 264: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 265: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 265: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 266: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 266: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 267: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 267: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 268: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 268: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 269: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 269: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 270: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 270: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 271: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 271: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 272: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 272: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 273: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 273: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 274: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 274: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 275: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 275: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 276: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 276: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 277: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 277: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 278: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 278: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 279: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 279: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 280: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 280: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 281: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 281: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 282: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 282: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 283: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 283: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 284: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 284: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 285: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 285: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 286: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 286: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 287: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 287: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 288: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 288: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 289: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 289: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 290: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 290: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 291: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 291: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 292: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 292: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 293: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 293: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 294: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 294: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 295: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 295: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0] Epoch: 296: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 296: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 297: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 297: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 298: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 298: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 299: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 299: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 100% 300/300 [1:15:11<00:00, 15.04s/it] INFO:root:Training is finished ... Iteration: 100% 24/24 [00:13<00:00, 1.73it/s, acc=1] Iteration: 100% 165/165 [23:50<00:00, 8.67s/it, acc=0.913] obj_pp_to_subj_pp: 16.8 cp_recursion: 55.5 pp_recursion: 52.9 subj_to_obj_proper: 98.5 prim_to_obj_proper: 95.8 prim_to_subj_proper: 99.9 LEX: 99.82666666666667 OVERALL: 91.27619047619048 INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 6046701 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.67 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:15<00:00, 14.19it/s, loss=5.76] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:14<00:00, 14.22it/s, loss=4.49] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:14<00:00, 14.26it/s, loss=3.44] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:14<00:00, 14.27it/s, loss=2.45] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:14<00:00, 14.28it/s, loss=1.89] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:14<00:00, 14.28it/s, loss=1.54] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:14<00:00, 14.28it/s, loss=1.28] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:14<00:00, 14.27it/s, loss=1.1] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.97] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.86] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.76] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0.68] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.61] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.52] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0.45] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.38] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.31] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.26] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0.22] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.19] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.17] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.15] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.12] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0.1] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.1] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.09] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.08] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.08] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.07] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0.07] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0.06] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.07] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0.06] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0.05] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0.06] Epoch: 35: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 35: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.04] Epoch: 36: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 36: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0.04] Epoch: 37: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 37: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0.04] Epoch: 38: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 38: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0.04] Epoch: 39: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 39: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0.03] Epoch: 40: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 40: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0.04] Epoch: 41: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 41: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.04] Epoch: 42: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 42: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0.03] Epoch: 43: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 43: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.03] Epoch: 44: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 44: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0.02] Epoch: 45: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 45: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0.03] Epoch: 46: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 46: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0.02] Epoch: 47: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 47: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.02] Epoch: 48: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 48: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.02] Epoch: 49: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 49: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0.02] Epoch: 50: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 50: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.02] Epoch: 51: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 51: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0.02] Epoch: 52: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 52: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.02] Epoch: 53: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 53: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.01] Epoch: 54: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 54: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.01] Epoch: 55: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 55: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0.01] Epoch: 56: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 56: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.01] Epoch: 57: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 57: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0.02] Epoch: 58: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 58: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.01] Epoch: 59: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 59: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0.01] Epoch: 60: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 60: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.01] Epoch: 61: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 61: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.01] Epoch: 62: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 62: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0.01] Epoch: 63: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 63: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0.01] Epoch: 64: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 64: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0.02] Epoch: 65: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 65: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.02] Epoch: 66: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 66: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.01] Epoch: 67: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 67: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0.01] Epoch: 68: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 68: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.01] Epoch: 69: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 69: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.01] Epoch: 70: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 70: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0.01] Epoch: 71: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 71: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0.01] Epoch: 72: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 72: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0.01] Epoch: 73: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 73: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0.01] Epoch: 74: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 74: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.01] Epoch: 75: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 75: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0.01] Epoch: 76: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 76: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.01] Epoch: 77: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 77: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0.01] Epoch: 78: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 78: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.01] Epoch: 79: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 79: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0.01] Epoch: 80: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 80: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0.01] Epoch: 81: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 81: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.01] Epoch: 82: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 82: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 83: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 83: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0.01] Epoch: 84: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 84: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.01] Epoch: 85: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 85: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.01] Epoch: 86: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 86: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 87: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 87: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 88: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 88: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0.01] Epoch: 89: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 89: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 90: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 90: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 91: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 91: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0.01] Epoch: 92: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 92: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0.01] Epoch: 93: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 93: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 94: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 94: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 95: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 95: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 96: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 96: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 97: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 97: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 98: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 98: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 99: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 99: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 100: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 100: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 101: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 101: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 102: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 102: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 103: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 103: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 104: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 104: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 105: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 105: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0.01] Epoch: 106: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 106: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 107: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 107: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 108: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 108: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 109: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 109: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 110: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 110: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 111: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 111: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 112: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 112: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 113: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 113: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 114: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 114: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 115: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 115: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 116: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 116: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 117: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 117: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 118: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 118: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 119: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 119: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 120: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 120: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 121: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 121: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 122: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 122: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 123: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 123: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 124: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 124: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 125: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 125: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 126: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 126: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 127: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 127: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 128: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 128: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 129: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 129: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 130: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 130: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 131: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 131: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 132: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 132: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 133: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 133: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 134: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 134: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 135: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 135: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 136: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 136: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 137: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 137: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 138: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 138: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 139: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 139: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 140: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 140: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 141: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 141: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 142: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 142: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 143: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 143: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 144: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 144: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 145: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 145: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 146: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 146: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 147: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 147: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 148: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 148: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 149: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 149: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 150: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 150: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 151: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 151: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 152: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 152: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 153: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 153: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 154: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 154: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 155: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 155: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 156: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 156: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 157: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 157: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 158: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 158: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 159: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 159: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 160: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 160: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 161: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 161: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 162: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 162: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 163: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 163: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 164: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 164: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 165: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 165: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 166: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 166: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 167: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 167: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 168: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 168: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 169: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 169: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 170: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 170: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 171: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 171: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 172: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 172: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 173: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 173: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 174: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 174: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 175: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 175: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 176: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 176: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 177: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 177: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 178: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 178: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 179: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 179: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 180: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 180: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 181: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 181: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 182: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 182: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 183: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 183: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 184: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 184: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 185: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 185: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 186: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 186: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 187: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 187: 100% 213/213 [00:15<00:00, 14.17it/s, loss=0] Epoch: 188: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 188: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 189: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 189: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 190: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 190: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 191: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 191: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 192: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 192: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 193: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 193: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 194: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 194: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 195: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 195: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 196: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 196: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 197: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 197: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 198: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 198: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 199: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 199: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 200: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 200: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 201: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 201: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 202: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 202: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 203: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 203: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 204: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 204: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 205: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 205: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 206: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 206: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 207: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 207: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 208: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 208: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 209: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 209: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 210: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 210: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 211: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 211: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 212: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 212: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 213: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 213: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 214: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 214: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 215: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 215: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 216: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 216: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 217: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 217: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 218: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 218: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 219: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 219: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 220: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 220: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 221: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 221: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 222: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 222: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 223: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 223: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 224: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 224: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 225: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 225: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 226: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 226: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 227: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 227: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 228: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 228: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 229: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 229: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 230: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 230: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 231: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 231: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 232: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 232: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 233: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 233: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 234: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 234: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 235: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 235: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 236: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 236: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 237: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 237: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 238: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 238: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 239: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 239: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 240: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 240: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 241: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 241: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 242: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 242: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 243: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 243: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 244: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 244: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 245: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 245: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 246: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 246: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 247: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 247: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 248: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 248: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 249: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 249: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 250: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 250: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 251: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 251: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 252: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 252: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 253: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 253: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 254: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 254: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 255: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 255: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 256: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 256: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 257: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 257: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 258: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 258: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 259: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 259: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 260: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 260: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 261: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 261: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 262: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 262: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 263: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 263: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 264: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 264: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 265: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 265: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 266: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 266: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 267: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 267: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 268: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 268: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 269: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 269: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 270: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 270: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 271: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 271: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 272: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 272: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 273: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 273: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 274: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 274: 1% 2/213 [00:00<00:11, 19.07it/s, loss=0]
colab interrupted this at seed 67
In [ ]:
!python run_cogs.py --model_name ende_transformer --gpu 1 --train_batch_size 128 --eval_batch_size 128 --lr 0.0001 --data_path ./recogs_positional_index --output_dir ./results_recogs_positional_index --lfs cogs --do_train --do_test --do_gen --max_seq_len 512 --output_json --epochs 300 --seeds "67;78;89;100" # paper's seeds, not sure how they chose them
INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 6046701 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.67 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]We strongly recommend passing in an `attention_mask` since your input_ids may be padded. See https://huggingface.co/docs/transformers/troubleshooting#incorrect-output-when-padding-tokens-arent-masked. /content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:17<00:00, 12.47it/s, loss=5.76] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:15<00:00, 13.99it/s, loss=4.49] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:15<00:00, 14.07it/s, loss=3.44] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:15<00:00, 14.05it/s, loss=2.45] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:15<00:00, 14.07it/s, loss=1.89] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:15<00:00, 14.07it/s, loss=1.54] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:15<00:00, 14.09it/s, loss=1.28] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:15<00:00, 14.07it/s, loss=1.1] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0.97] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:15<00:00, 14.14it/s, loss=0.86] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:15<00:00, 14.09it/s, loss=0.76] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:15<00:00, 14.15it/s, loss=0.68] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0.61] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0.52] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0.45] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0.38] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0.31] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0.26] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.22] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0.19] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.17] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0.15] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.12] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0.1] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.1] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.09] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0.08] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.08] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.07] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0.07] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.06] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.07] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.06] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.05] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.06] Epoch: 35: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 35: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.04] Epoch: 36: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 36: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0.04] Epoch: 37: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 37: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0.04] Epoch: 38: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 38: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.04] Epoch: 39: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 39: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.03] Epoch: 40: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 40: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.04] Epoch: 41: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 41: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.04] Epoch: 42: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 42: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.03] Epoch: 43: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 43: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.03] Epoch: 44: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 44: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.02] Epoch: 45: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 45: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.03] Epoch: 46: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 46: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0.02] Epoch: 47: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 47: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.02] Epoch: 48: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 48: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.02] Epoch: 49: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 49: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.02] Epoch: 50: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 50: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.02] Epoch: 51: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 51: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.02] Epoch: 52: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 52: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.02] Epoch: 53: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 53: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.01] Epoch: 54: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 54: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0.01] Epoch: 55: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 55: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.01] Epoch: 56: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 56: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0.01] Epoch: 57: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 57: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.02] Epoch: 58: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 58: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.01] Epoch: 59: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 59: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.01] Epoch: 60: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 60: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.01] Epoch: 61: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 61: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.01] Epoch: 62: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 62: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.01] Epoch: 63: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 63: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.01] Epoch: 64: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 64: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.02] Epoch: 65: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 65: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.02] Epoch: 66: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 66: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0.01] Epoch: 67: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 67: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.01] Epoch: 68: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 68: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.01] Epoch: 69: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 69: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.01] Epoch: 70: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 70: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0.01] Epoch: 71: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 71: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.01] Epoch: 72: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 72: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.01] Epoch: 73: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 73: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.01] Epoch: 74: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 74: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.01] Epoch: 75: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 75: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.01] Epoch: 76: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 76: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.01] Epoch: 77: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 77: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.01] Epoch: 78: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 78: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.01] Epoch: 79: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 79: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0.01] Epoch: 80: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 80: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.01] Epoch: 81: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 81: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.01] Epoch: 82: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 82: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 83: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 83: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.01] Epoch: 84: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 84: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.01] Epoch: 85: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 85: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.01] Epoch: 86: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 86: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 87: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 87: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 88: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 88: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.01] Epoch: 89: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 89: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 90: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 90: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 91: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 91: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0.01] Epoch: 92: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 92: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.01] Epoch: 93: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 93: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 94: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 94: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 95: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 95: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 96: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 96: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 97: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 97: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 98: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 98: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 99: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 99: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 100: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 100: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 101: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 101: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 102: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 102: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 103: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 103: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 104: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 104: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 105: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 105: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0.01] Epoch: 106: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 106: 100% 213/213 [00:15<00:00, 14.03it/s, loss=0] Epoch: 107: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 107: 100% 213/213 [00:15<00:00, 14.01it/s, loss=0] Epoch: 108: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 108: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 109: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 109: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 110: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 110: 100% 213/213 [00:15<00:00, 14.04it/s, loss=0] Epoch: 111: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 111: 100% 213/213 [00:15<00:00, 14.01it/s, loss=0] Epoch: 112: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 112: 100% 213/213 [00:15<00:00, 14.03it/s, loss=0] Epoch: 113: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 113: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 114: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 114: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 115: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 115: 100% 213/213 [00:15<00:00, 13.99it/s, loss=0] Epoch: 116: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 116: 100% 213/213 [00:15<00:00, 14.03it/s, loss=0] Epoch: 117: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 117: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 118: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 118: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 119: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 119: 100% 213/213 [00:15<00:00, 14.03it/s, loss=0] Epoch: 120: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 120: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 121: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 121: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 122: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 122: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 123: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 123: 100% 213/213 [00:15<00:00, 14.01it/s, loss=0] Epoch: 124: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 124: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 125: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 125: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 126: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 126: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 127: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 127: 100% 213/213 [00:15<00:00, 14.03it/s, loss=0] Epoch: 128: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 128: 100% 213/213 [00:15<00:00, 14.03it/s, loss=0] Epoch: 129: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 129: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 130: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 130: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 131: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 131: 100% 213/213 [00:15<00:00, 14.04it/s, loss=0] Epoch: 132: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 132: 100% 213/213 [00:15<00:00, 14.04it/s, loss=0] Epoch: 133: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 133: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 134: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 134: 100% 213/213 [00:15<00:00, 14.03it/s, loss=0] Epoch: 135: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 135: 100% 213/213 [00:15<00:00, 13.99it/s, loss=0] Epoch: 136: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 136: 100% 213/213 [00:15<00:00, 14.04it/s, loss=0] Epoch: 137: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 137: 100% 213/213 [00:15<00:00, 14.03it/s, loss=0] Epoch: 138: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 138: 100% 213/213 [00:15<00:00, 14.04it/s, loss=0] Epoch: 139: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 139: 100% 213/213 [00:15<00:00, 14.01it/s, loss=0] Epoch: 140: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 140: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 141: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 141: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 142: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 142: 100% 213/213 [00:15<00:00, 14.04it/s, loss=0] Epoch: 143: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 143: 100% 213/213 [00:15<00:00, 14.01it/s, loss=0] Epoch: 144: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 144: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 145: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 145: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 146: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 146: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 147: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 147: 100% 213/213 [00:15<00:00, 14.03it/s, loss=0] Epoch: 148: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 148: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 149: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 149: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 150: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 150: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 151: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 151: 100% 213/213 [00:15<00:00, 14.03it/s, loss=0] Epoch: 152: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 152: 100% 213/213 [00:15<00:00, 14.04it/s, loss=0] Epoch: 153: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 153: 100% 213/213 [00:15<00:00, 14.09it/s, loss=0] Epoch: 154: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 154: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0] Epoch: 155: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 155: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 156: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 156: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 157: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 157: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 158: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 158: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 159: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 159: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 160: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 160: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0] Epoch: 161: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 161: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 162: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 162: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 163: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 163: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 164: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 164: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 165: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 165: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 166: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 166: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 167: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 167: 100% 213/213 [00:15<00:00, 14.15it/s, loss=0] Epoch: 168: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 168: 100% 213/213 [00:15<00:00, 14.15it/s, loss=0] Epoch: 169: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 169: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 170: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 170: 100% 213/213 [00:15<00:00, 14.17it/s, loss=0] Epoch: 171: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 171: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0] Epoch: 172: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 172: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0] Epoch: 173: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 173: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 174: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 174: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 175: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 175: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 176: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 176: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0] Epoch: 177: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 177: 100% 213/213 [00:15<00:00, 14.17it/s, loss=0] Epoch: 178: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 178: 100% 213/213 [00:15<00:00, 14.17it/s, loss=0] Epoch: 179: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 179: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 180: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 180: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0] Epoch: 181: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 181: 100% 213/213 [00:15<00:00, 14.14it/s, loss=0] Epoch: 182: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 182: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 183: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 183: 100% 213/213 [00:15<00:00, 14.10it/s, loss=0] Epoch: 184: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 184: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 185: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 185: 100% 213/213 [00:15<00:00, 14.09it/s, loss=0] Epoch: 186: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 186: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 187: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 187: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 188: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 188: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 189: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 189: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 190: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 190: 100% 213/213 [00:15<00:00, 14.11it/s, loss=0] Epoch: 191: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 191: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 192: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 192: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0] Epoch: 193: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 193: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 194: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 194: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0] Epoch: 195: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 195: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 196: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 196: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 197: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 197: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 198: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 198: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 199: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 199: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 200: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 200: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 201: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 201: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 202: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 202: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 203: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 203: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 204: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 204: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 205: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 205: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 206: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 206: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 207: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 207: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 208: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 208: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 209: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 209: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 210: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 210: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 211: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 211: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 212: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 212: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 213: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 213: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 214: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 214: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 215: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 215: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 216: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 216: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 217: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 217: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 218: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 218: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 219: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 219: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 220: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 220: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 221: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 221: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 222: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 222: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 223: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 223: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 224: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 224: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 225: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 225: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 226: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 226: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 227: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 227: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 228: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 228: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 229: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 229: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 230: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 230: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 231: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 231: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 232: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 232: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 233: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 233: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 234: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 234: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 235: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 235: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 236: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 236: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 237: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 237: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 238: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 238: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 239: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 239: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 240: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 240: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 241: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 241: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 242: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 242: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 243: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 243: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 244: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 244: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 245: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 245: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 246: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 246: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 247: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 247: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 248: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 248: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 249: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 249: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 250: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 250: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 251: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 251: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 252: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 252: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 253: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 253: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 254: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 254: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 255: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 255: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 256: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 256: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 257: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 257: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 258: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 258: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 259: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 259: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 260: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 260: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 261: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 261: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 262: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 262: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 263: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 263: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 264: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 264: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 265: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 265: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 266: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 266: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 267: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 267: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 268: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 268: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 269: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 269: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 270: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 270: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 271: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 271: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 272: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 272: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 273: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 273: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 274: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 274: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 275: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 275: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 276: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 276: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 277: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 277: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 278: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 278: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 279: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 279: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 280: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 280: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 281: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 281: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 282: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 282: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 283: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 283: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 284: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 284: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 285: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 285: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 286: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 286: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 287: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 287: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 288: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 288: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 289: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 289: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 290: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 290: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 291: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 291: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 292: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 292: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 293: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 293: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 294: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 294: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 295: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 295: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 296: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 296: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 297: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 297: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 298: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 298: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 299: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 299: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 100% 300/300 [1:15:20<00:00, 15.07s/it] INFO:root:Training is finished ... Iteration: 100% 24/24 [00:13<00:00, 1.72it/s, acc=1] Iteration: 100% 165/165 [34:43<00:00, 12.62s/it, acc=0.871] obj_pp_to_subj_pp: 18.1 cp_recursion: 56.5 pp_recursion: 23.9 subj_to_obj_proper: 81.0 prim_to_obj_proper: 84.0 prim_to_subj_proper: 100.0 LEX: 97.70666666666666 OVERALL: 87.1 INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 6046701 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.78 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:15<00:00, 14.11it/s, loss=5.76] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:14<00:00, 14.22it/s, loss=4.53] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:15<00:00, 14.18it/s, loss=3.48] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:14<00:00, 14.22it/s, loss=2.42] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:14<00:00, 14.22it/s, loss=1.86] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:14<00:00, 14.22it/s, loss=1.5] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:15<00:00, 14.18it/s, loss=1.25] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:14<00:00, 14.21it/s, loss=1.08] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0.95] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:15<00:00, 14.15it/s, loss=0.84] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0.74] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:15<00:00, 14.12it/s, loss=0.66] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:15<00:00, 14.11it/s, loss=0.58] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:15<00:00, 14.10it/s, loss=0.49] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:15<00:00, 14.09it/s, loss=0.43] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:15<00:00, 14.09it/s, loss=0.36] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0.31] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:15<00:00, 14.10it/s, loss=0.26] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:15<00:00, 14.14it/s, loss=0.21] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0.18] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0.15] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0.12] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0.12] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0.09] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.08] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.06] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.06] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.05] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.04] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.04] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.04] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.03] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.03] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0.03] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.03] Epoch: 35: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 35: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.02] Epoch: 36: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 36: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0.02] Epoch: 37: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 37: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0.02] Epoch: 38: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 38: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0.02] Epoch: 39: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 39: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0.01] Epoch: 40: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 40: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0.02] Epoch: 41: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 41: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0.01] Epoch: 42: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 42: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.01] Epoch: 43: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 43: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0.01] Epoch: 44: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 44: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0.01] Epoch: 45: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 45: 100% 213/213 [00:14<00:00, 14.37it/s, loss=0.01] Epoch: 46: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 46: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.01] Epoch: 47: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 47: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0.01] Epoch: 48: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 48: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.01] Epoch: 49: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 49: 100% 213/213 [00:14<00:00, 14.37it/s, loss=0] Epoch: 50: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 50: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 51: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 51: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 52: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 52: 100% 213/213 [00:14<00:00, 14.37it/s, loss=0.01] Epoch: 53: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 53: 100% 213/213 [00:14<00:00, 14.37it/s, loss=0.01] Epoch: 54: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 54: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 55: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 55: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.01] Epoch: 56: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 56: 100% 213/213 [00:14<00:00, 14.38it/s, loss=0] Epoch: 57: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 57: 100% 213/213 [00:14<00:00, 14.37it/s, loss=0] Epoch: 58: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 58: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 59: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 59: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 60: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 60: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 61: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 61: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 62: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 62: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 63: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 63: 100% 213/213 [00:14<00:00, 14.37it/s, loss=0] Epoch: 64: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 64: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 65: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 65: 100% 213/213 [00:14<00:00, 14.37it/s, loss=0] Epoch: 66: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 66: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 67: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 67: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 68: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 68: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 69: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 69: 100% 213/213 [00:14<00:00, 14.37it/s, loss=0] Epoch: 70: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 70: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 71: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 71: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 72: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 72: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 73: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 73: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 74: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 74: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.01] Epoch: 75: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 75: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 76: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 76: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 77: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 77: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 78: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 78: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 79: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 79: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 80: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 80: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 81: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 81: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 82: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 82: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 83: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 83: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 84: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 84: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 85: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 85: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 86: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 86: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 87: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 87: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 88: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 88: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 89: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 89: 100% 213/213 [00:15<00:00, 14.09it/s, loss=0] Epoch: 90: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 90: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 91: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 91: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 92: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 92: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 93: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 93: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 94: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 94: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 95: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 95: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 96: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 96: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 97: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 97: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 98: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 98: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 99: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 99: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 100: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 100: 100% 213/213 [00:15<00:00, 14.03it/s, loss=0] Epoch: 101: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 101: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 102: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 102: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 103: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 103: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 104: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 104: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 105: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 105: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 106: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 106: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 107: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 107: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 108: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 108: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 109: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 109: 100% 213/213 [00:15<00:00, 14.10it/s, loss=0] Epoch: 110: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 110: 100% 213/213 [00:15<00:00, 14.09it/s, loss=0] Epoch: 111: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 111: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 112: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 112: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 113: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 113: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 114: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 114: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 115: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 115: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 116: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 116: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 117: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 117: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 118: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 118: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 119: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 119: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 120: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 120: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0.01] Epoch: 121: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 121: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 122: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 122: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 123: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 123: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 124: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 124: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 125: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 125: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 126: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 126: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0] Epoch: 127: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 127: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 128: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 128: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 129: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 129: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 130: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 130: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 131: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 131: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 132: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 132: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 133: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 133: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 134: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 134: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0] Epoch: 135: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 135: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 136: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 136: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 137: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 137: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 138: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 138: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 139: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 139: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 140: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 140: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 141: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 141: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 142: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 142: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 143: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 143: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 144: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 144: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 145: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 145: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 146: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 146: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 147: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 147: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 148: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 148: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 149: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 149: 100% 213/213 [00:15<00:00, 14.17it/s, loss=0] Epoch: 150: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 150: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 151: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 151: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 152: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 152: 100% 213/213 [00:15<00:00, 14.17it/s, loss=0] Epoch: 153: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 153: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 154: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 154: 100% 213/213 [00:15<00:00, 14.12it/s, loss=0] Epoch: 155: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 155: 100% 213/213 [00:15<00:00, 14.11it/s, loss=0] Epoch: 156: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 156: 100% 213/213 [00:15<00:00, 14.10it/s, loss=0] Epoch: 157: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 157: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 158: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 158: 100% 213/213 [00:15<00:00, 14.09it/s, loss=0] Epoch: 159: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 159: 100% 213/213 [00:15<00:00, 14.11it/s, loss=0] Epoch: 160: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 160: 100% 213/213 [00:15<00:00, 14.09it/s, loss=0] Epoch: 161: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 161: 100% 213/213 [00:15<00:00, 14.10it/s, loss=0] Epoch: 162: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 162: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 163: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 163: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 164: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 164: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 165: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 165: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0] Epoch: 166: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 166: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 167: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 167: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 168: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 168: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 169: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 169: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 170: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 170: 100% 213/213 [00:14<00:00, 14.37it/s, loss=0] Epoch: 171: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 171: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 172: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 172: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 173: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 173: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 174: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 174: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 175: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 175: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 176: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 176: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 177: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 177: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 178: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 178: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 179: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 179: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 180: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 180: 100% 213/213 [00:14<00:00, 14.37it/s, loss=0] Epoch: 181: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 181: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 182: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 182: 100% 213/213 [00:14<00:00, 14.37it/s, loss=0] Epoch: 183: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 183: 100% 213/213 [00:14<00:00, 14.37it/s, loss=0] Epoch: 184: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 184: 100% 213/213 [00:14<00:00, 14.37it/s, loss=0] Epoch: 185: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 185: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 186: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 186: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 187: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 187: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 188: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 188: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 189: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 189: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 190: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 190: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 191: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 191: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 192: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 192: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 193: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 193: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 194: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 194: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 195: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 195: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 196: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 196: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 197: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 197: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 198: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 198: 100% 213/213 [00:15<00:00, 14.03it/s, loss=0] Epoch: 199: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 199: 100% 213/213 [00:15<00:00, 14.09it/s, loss=0] Epoch: 200: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 200: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 201: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 201: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 202: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 202: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 203: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 203: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 204: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 204: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 205: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 205: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 206: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 206: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 207: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 207: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 208: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 208: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 209: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 209: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 210: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 210: 100% 213/213 [00:15<00:00, 14.02it/s, loss=0] Epoch: 211: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 211: 100% 213/213 [00:15<00:00, 14.04it/s, loss=0] Epoch: 212: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 212: 100% 213/213 [00:15<00:00, 14.04it/s, loss=0] Epoch: 213: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 213: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 214: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 214: 100% 213/213 [00:15<00:00, 14.15it/s, loss=0] Epoch: 215: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 215: 100% 213/213 [00:14<00:00, 14.37it/s, loss=0] Epoch: 216: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 216: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 217: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 217: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 218: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 218: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 219: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 219: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 220: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 220: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 221: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 221: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 222: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 222: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 223: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 223: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 224: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 224: 100% 213/213 [00:15<00:00, 14.09it/s, loss=0] Epoch: 225: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 225: 100% 213/213 [00:15<00:00, 14.09it/s, loss=0] Epoch: 226: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 226: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 227: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 227: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 228: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 228: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 229: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 229: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 230: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 230: 100% 213/213 [00:15<00:00, 14.04it/s, loss=0] Epoch: 231: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 231: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 232: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 232: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 233: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 233: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 234: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 234: 100% 213/213 [00:15<00:00, 14.04it/s, loss=0] Epoch: 235: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 235: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 236: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 236: 100% 213/213 [00:15<00:00, 14.09it/s, loss=0] Epoch: 237: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 237: 100% 213/213 [00:15<00:00, 14.09it/s, loss=0] Epoch: 238: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 238: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 239: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 239: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 240: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 240: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 241: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 241: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 242: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 242: 100% 213/213 [00:15<00:00, 14.02it/s, loss=0] Epoch: 243: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 243: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 244: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 244: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 245: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 245: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 246: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 246: 100% 213/213 [00:15<00:00, 14.02it/s, loss=0] Epoch: 247: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 247: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 248: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 248: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 249: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 249: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 250: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 250: 100% 213/213 [00:15<00:00, 14.03it/s, loss=0] Epoch: 251: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 251: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 252: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 252: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 253: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 253: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 254: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 254: 100% 213/213 [00:15<00:00, 14.02it/s, loss=0] Epoch: 255: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 255: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 256: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 256: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 257: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 257: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 258: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 258: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 259: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 259: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 260: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 260: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 261: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 261: 100% 213/213 [00:15<00:00, 14.09it/s, loss=0] Epoch: 262: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 262: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 263: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 263: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 264: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 264: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 265: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 265: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 266: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 266: 100% 213/213 [00:15<00:00, 14.02it/s, loss=0] Epoch: 267: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 267: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 268: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 268: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 269: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 269: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 270: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 270: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 271: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 271: 100% 213/213 [00:15<00:00, 14.13it/s, loss=0] Epoch: 272: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 272: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 273: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 273: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 274: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 274: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 275: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 275: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 276: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 276: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 277: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 277: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 278: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 278: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 279: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 279: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0] Epoch: 280: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 280: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 281: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 281: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 282: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 282: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 283: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 283: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 284: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 284: 100% 213/213 [00:15<00:00, 14.17it/s, loss=0] Epoch: 285: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 285: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 286: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 286: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0] Epoch: 287: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 287: 100% 213/213 [00:15<00:00, 14.17it/s, loss=0] Epoch: 288: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 288: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 289: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 289: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 290: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 290: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 291: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 291: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0] Epoch: 292: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 292: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 293: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 293: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 294: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 294: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 295: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 295: 100% 213/213 [00:15<00:00, 14.17it/s, loss=0] Epoch: 296: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 296: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 297: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 297: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 298: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 298: 100% 213/213 [00:15<00:00, 14.15it/s, loss=0] Epoch: 299: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 299: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 100% 300/300 [1:15:31<00:00, 15.10s/it] INFO:root:Training is finished ... Iteration: 100% 24/24 [00:14<00:00, 1.66it/s, acc=1] Iteration: 100% 165/165 [29:31<00:00, 10.73s/it, acc=0.903] obj_pp_to_subj_pp: 17.5 cp_recursion: 58.9 pp_recursion: 50.1 subj_to_obj_proper: 85.2 prim_to_obj_proper: 86.5 prim_to_subj_proper: 99.7 LEX: 99.93333333333332 OVERALL: 90.32857142857142 INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 6046701 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.89 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:15<00:00, 14.14it/s, loss=5.79] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:15<00:00, 14.20it/s, loss=4.49] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:15<00:00, 14.20it/s, loss=3.54] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:15<00:00, 14.17it/s, loss=2.48] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:15<00:00, 14.20it/s, loss=1.91] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:15<00:00, 14.18it/s, loss=1.55] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:15<00:00, 14.19it/s, loss=1.29] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:15<00:00, 14.18it/s, loss=1.11] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0.98] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0.87] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0.77] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0.67] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:15<00:00, 14.12it/s, loss=0.58] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:15<00:00, 14.09it/s, loss=0.5] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:15<00:00, 14.11it/s, loss=0.43] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:15<00:00, 14.02it/s, loss=0.37] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:15<00:00, 14.11it/s, loss=0.32] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:15<00:00, 14.09it/s, loss=0.28] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:15<00:00, 14.10it/s, loss=0.24] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0.19] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:15<00:00, 14.10it/s, loss=0.17] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:15<00:00, 14.14it/s, loss=0.15] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0.12] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0.12] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0.1] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0.09] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0.09] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.09] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0.08] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0.07] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.05] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.06] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.04] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0.04] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.05] Epoch: 35: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 35: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.04] Epoch: 36: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 36: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0.04] Epoch: 37: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 37: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0.03] Epoch: 38: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 38: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0.03] Epoch: 39: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 39: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.03] Epoch: 40: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 40: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.02] Epoch: 41: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 41: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.02] Epoch: 42: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 42: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.03] Epoch: 43: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 43: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.02] Epoch: 44: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 44: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.02] Epoch: 45: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 45: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.02] Epoch: 46: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 46: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.02] Epoch: 47: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 47: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0.02] Epoch: 48: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 48: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.02] Epoch: 49: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 49: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.02] Epoch: 50: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 50: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.01] Epoch: 51: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 51: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.01] Epoch: 52: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 52: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.01] Epoch: 53: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 53: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0.01] Epoch: 54: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 54: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.01] Epoch: 55: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 55: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.01] Epoch: 56: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 56: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.01] Epoch: 57: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 57: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.01] Epoch: 58: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 58: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.01] Epoch: 59: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 59: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.01] Epoch: 60: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 60: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0.01] Epoch: 61: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 61: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 62: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 62: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0.01] Epoch: 63: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 63: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 64: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 64: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.01] Epoch: 65: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 65: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.01] Epoch: 66: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 66: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.01] Epoch: 67: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 67: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.01] Epoch: 68: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 68: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.01] Epoch: 69: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 69: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.01] Epoch: 70: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 70: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 71: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 71: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.01] Epoch: 72: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 72: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.01] Epoch: 73: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 73: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0.01] Epoch: 74: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 74: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0.01] Epoch: 75: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 75: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 76: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 76: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 77: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 77: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 78: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 78: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0.01] Epoch: 79: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 79: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 80: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 80: 100% 213/213 [00:15<00:00, 14.03it/s, loss=0] Epoch: 81: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 81: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 82: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 82: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 83: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 83: 100% 213/213 [00:15<00:00, 14.04it/s, loss=0] Epoch: 84: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 84: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 85: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 85: 100% 213/213 [00:15<00:00, 14.09it/s, loss=0] Epoch: 86: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 86: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 87: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 87: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 88: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 88: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 89: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 89: 100% 213/213 [00:15<00:00, 14.03it/s, loss=0.01] Epoch: 90: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 90: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0.01] Epoch: 91: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 91: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 92: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 92: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 93: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 93: 100% 213/213 [00:15<00:00, 14.04it/s, loss=0] Epoch: 94: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 94: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0.01] Epoch: 95: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 95: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0.01] Epoch: 96: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 96: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 97: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 97: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 98: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 98: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 99: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 99: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 100: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 100: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 101: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 101: 100% 213/213 [00:15<00:00, 14.02it/s, loss=0] Epoch: 102: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 102: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 103: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 103: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 104: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 104: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0.01] Epoch: 105: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 105: 100% 213/213 [00:15<00:00, 14.03it/s, loss=0] Epoch: 106: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 106: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 107: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 107: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 108: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 108: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 109: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 109: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 110: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 110: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 111: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 111: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 112: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 112: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 113: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 113: 100% 213/213 [00:15<00:00, 14.03it/s, loss=0] Epoch: 114: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 114: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 115: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 115: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 116: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 116: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 117: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 117: 100% 213/213 [00:15<00:00, 14.03it/s, loss=0] Epoch: 118: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 118: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 119: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 119: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 120: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 120: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 121: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 121: 100% 213/213 [00:15<00:00, 14.04it/s, loss=0] Epoch: 122: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 122: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 123: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 123: 100% 213/213 [00:15<00:00, 14.04it/s, loss=0] Epoch: 124: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 124: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 125: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 125: 100% 213/213 [00:15<00:00, 14.04it/s, loss=0] Epoch: 126: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 126: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 127: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 127: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 128: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 128: 100% 213/213 [00:15<00:00, 14.05it/s, loss=0] Epoch: 129: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 129: 100% 213/213 [00:15<00:00, 14.04it/s, loss=0] Epoch: 130: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 130: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 131: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 131: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 132: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 132: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 133: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 133: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 134: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 134: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 135: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 135: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 136: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 136: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 137: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 137: 100% 213/213 [00:15<00:00, 14.03it/s, loss=0] Epoch: 138: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 138: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 139: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 139: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 140: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 140: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 141: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 141: 100% 213/213 [00:15<00:00, 14.04it/s, loss=0] Epoch: 142: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 142: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 143: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 143: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 144: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 144: 100% 213/213 [00:15<00:00, 14.04it/s, loss=0] Epoch: 145: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 145: 100% 213/213 [00:15<00:00, 14.02it/s, loss=0] Epoch: 146: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 146: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 147: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 147: 100% 213/213 [00:15<00:00, 14.15it/s, loss=0] Epoch: 148: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 148: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 149: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 149: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 150: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 150: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 151: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 151: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 152: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 152: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 153: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 153: 100% 213/213 [00:14<00:00, 14.21it/s, loss=0] Epoch: 154: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 154: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 155: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 155: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 156: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 156: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 157: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 157: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 158: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 158: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 159: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 159: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 160: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 160: 100% 213/213 [00:15<00:00, 14.14it/s, loss=0] Epoch: 161: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 161: 100% 213/213 [00:15<00:00, 14.15it/s, loss=0] Epoch: 162: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 162: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 163: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 163: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 164: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 164: 100% 213/213 [00:15<00:00, 14.20it/s, loss=0] Epoch: 165: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 165: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 166: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 166: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 167: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 167: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 168: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 168: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 169: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 169: 100% 213/213 [00:15<00:00, 14.14it/s, loss=0] Epoch: 170: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 170: 100% 213/213 [00:15<00:00, 14.17it/s, loss=0] Epoch: 171: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 171: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 172: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 172: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 173: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 173: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 174: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 174: 100% 213/213 [00:15<00:00, 14.11it/s, loss=0] Epoch: 175: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 175: 100% 213/213 [00:15<00:00, 14.10it/s, loss=0] Epoch: 176: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 176: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 177: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 177: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 178: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 178: 100% 213/213 [00:15<00:00, 14.07it/s, loss=0] Epoch: 179: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 179: 100% 213/213 [00:15<00:00, 14.08it/s, loss=0] Epoch: 180: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 180: 100% 213/213 [00:15<00:00, 14.09it/s, loss=0] Epoch: 181: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 181: 100% 213/213 [00:15<00:00, 14.06it/s, loss=0] Epoch: 182: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 182: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 183: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 183: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 184: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 184: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 185: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 185: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 186: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 186: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 187: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 187: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 188: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 188: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 189: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 189: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 190: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 190: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 191: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 191: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 192: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 192: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 193: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 193: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 194: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 194: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 195: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 195: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 196: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 196: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 197: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 197: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 198: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 198: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 199: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 199: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 200: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 200: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 201: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 201: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 202: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 202: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 203: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 203: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 204: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 204: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 205: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 205: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 206: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 206: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 207: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 207: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 208: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 208: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 209: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 209: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 210: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 210: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 211: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 211: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 212: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 212: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 213: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 213: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 214: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 214: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 215: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 215: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 216: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 216: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 217: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 217: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 218: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 218: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 219: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 219: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 220: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 220: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 221: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 221: 100% 213/213 [00:14<00:00, 14.37it/s, loss=0] Epoch: 222: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 222: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 223: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 223: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 224: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 224: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 225: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 225: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 226: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 226: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 227: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 227: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 228: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 228: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 229: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 229: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 230: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 230: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 231: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 231: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 232: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 232: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 233: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 233: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 234: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 234: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 235: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 235: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 236: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 236: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 237: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 237: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 238: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 238: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 239: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 239: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 240: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 240: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 241: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 241: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 242: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 242: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 243: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 243: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 244: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 244: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 245: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 245: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 246: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 246: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 247: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 247: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 248: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 248: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 249: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 249: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 250: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 250: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 251: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 251: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 252: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 252: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 253: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 253: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 254: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 254: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 255: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 255: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 256: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 256: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 257: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 257: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 258: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 258: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 259: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 259: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 260: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 260: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 261: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 261: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 262: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 262: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 263: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 263: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 264: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 264: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 265: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 265: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 266: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 266: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 267: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 267: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 268: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 268: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 269: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 269: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 270: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 270: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 271: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 271: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 272: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 272: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 273: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 273: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 274: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 274: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 275: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 275: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 276: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 276: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 277: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 277: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 278: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 278: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 279: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 279: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 280: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 280: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 281: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 281: 100% 213/213 [00:14<00:00, 14.36it/s, loss=0] Epoch: 282: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 282: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 283: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 283: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 284: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 284: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 285: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 285: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 286: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 286: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 287: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 287: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 288: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 288: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 289: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 289: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 290: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 290: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 291: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 291: 100% 213/213 [00:14<00:00, 14.34it/s, loss=0] Epoch: 292: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 292: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 293: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 293: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 294: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 294: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 295: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 295: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 296: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 296: 100% 213/213 [00:14<00:00, 14.35it/s, loss=0] Epoch: 297: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 297: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 298: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 298: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 299: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 299: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 100% 300/300 [1:15:17<00:00, 15.06s/it] INFO:root:Training is finished ... Iteration: 100% 24/24 [00:14<00:00, 1.68it/s, acc=1] Iteration: 100% 165/165 [25:10<00:00, 9.16s/it, acc=0.887] obj_pp_to_subj_pp: 21.5 cp_recursion: 58.4 pp_recursion: 54.2 subj_to_obj_proper: 93.0 prim_to_obj_proper: 46.3 prim_to_subj_proper: 98.9 LEX: 99.42666666666666 OVERALL: 88.74761904761905 INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 6046701 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.100 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:14<00:00, 14.22it/s, loss=5.85] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:14<00:00, 14.24it/s, loss=4.56] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:14<00:00, 14.25it/s, loss=3.52] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:14<00:00, 14.26it/s, loss=2.45] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:15<00:00, 14.20it/s, loss=1.88] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:15<00:00, 14.19it/s, loss=1.53] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:14<00:00, 14.29it/s, loss=1.26] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:14<00:00, 14.26it/s, loss=1.08] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.96] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0.86] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.77] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0.68] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.58] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.49] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0.41] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.35] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.28] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.23] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0.18] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.15] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.11] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.09] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0.08] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.07] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.05] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.05] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.06] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.04] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.05] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.04] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.04] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.04] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.02] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.02] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.02] Epoch: 35: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 35: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.02] Epoch: 36: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 36: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.02] Epoch: 37: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 37: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.02] Epoch: 38: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 38: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.02] Epoch: 39: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 39: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.01] Epoch: 40: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 40: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.01] Epoch: 41: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 41: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0.01] Epoch: 42: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 42: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0.01] Epoch: 43: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 43: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.01] Epoch: 44: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 44: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.01] Epoch: 45: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 45: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.01] Epoch: 46: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 46: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0.01] Epoch: 47: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 47: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0.01] Epoch: 48: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 48: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.01] Epoch: 49: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 49: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0.01] Epoch: 50: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 50: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 51: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 51: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.01] Epoch: 52: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 52: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 53: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 53: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.01] Epoch: 54: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 54: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.01] Epoch: 55: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 55: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0.01] Epoch: 56: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 56: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 57: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 57: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 58: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 58: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 59: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 59: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 60: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 60: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.01] Epoch: 61: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 61: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0.01] Epoch: 62: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 62: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 63: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 63: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 64: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 64: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 65: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 65: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 66: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 66: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 67: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 67: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 68: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 68: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 69: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 69: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 70: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 70: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 71: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 71: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 72: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 72: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 73: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 73: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 74: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 74: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 75: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 75: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 76: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 76: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0.01] Epoch: 77: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 77: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 78: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 78: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 79: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 79: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 80: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 80: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 81: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 81: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 82: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 82: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 83: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 83: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 84: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 84: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 85: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 85: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 86: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 86: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 87: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 87: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 88: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 88: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 89: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 89: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 90: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 90: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 91: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 91: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 92: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 92: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 93: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 93: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 94: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 94: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 95: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 95: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 96: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 96: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 97: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 97: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 98: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 98: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 99: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 99: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 100: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 100: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 101: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 101: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 102: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 102: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 103: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 103: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 104: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 104: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 105: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 105: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 106: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 106: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 107: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 107: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 108: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 108: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 109: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 109: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 110: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 110: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 111: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 111: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 112: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 112: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 113: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 113: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 114: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 114: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 115: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 115: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 116: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 116: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 117: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 117: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 118: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 118: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 119: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 119: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 120: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 120: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 121: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 121: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 122: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 122: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 123: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 123: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 124: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 124: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 125: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 125: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 126: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 126: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 127: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 127: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 128: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 128: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 129: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 129: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 130: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 130: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 131: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 131: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 132: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 132: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 133: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 133: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 134: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 134: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 135: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 135: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 136: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 136: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 137: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 137: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 138: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 138: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 139: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 139: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 140: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 140: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 141: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 141: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 142: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 142: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 143: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 143: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 144: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 144: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 145: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 145: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 146: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 146: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 147: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 147: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 148: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 148: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 149: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 149: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 150: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 150: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 151: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 151: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 152: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 152: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 153: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 153: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 154: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 154: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 155: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 155: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 156: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 156: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 157: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 157: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 158: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 158: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 159: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 159: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 160: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 160: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 161: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 161: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 162: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 162: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 163: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 163: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 164: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 164: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 165: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 165: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 166: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 166: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 167: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 167: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 168: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 168: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 169: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 169: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 170: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 170: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 171: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 171: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 172: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 172: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 173: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 173: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 174: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 174: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 175: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 175: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 176: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 176: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 177: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 177: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 178: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 178: 100% 213/213 [00:15<00:00, 14.16it/s, loss=0] Epoch: 179: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 179: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 180: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 180: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 181: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 181: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 182: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 182: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 183: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 183: 100% 213/213 [00:14<00:00, 14.33it/s, loss=0] Epoch: 184: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 184: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 185: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 185: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 186: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 186: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 187: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 187: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 188: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 188: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 189: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 189: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 190: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 190: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 191: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 191: 100% 213/213 [00:15<00:00, 14.17it/s, loss=0] Epoch: 192: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 192: 100% 213/213 [00:15<00:00, 14.18it/s, loss=0] Epoch: 193: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 193: 100% 213/213 [00:15<00:00, 14.19it/s, loss=0] Epoch: 194: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 194: 100% 213/213 [00:15<00:00, 14.17it/s, loss=0] Epoch: 195: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 195: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 196: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 196: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 197: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 197: 100% 213/213 [00:14<00:00, 14.20it/s, loss=0] Epoch: 198: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 198: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 199: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 199: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 200: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 200: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 201: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 201: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 202: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 202: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 203: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 203: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 204: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 204: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 205: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 205: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 206: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 206: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 207: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 207: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 208: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 208: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 209: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 209: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 210: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 210: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 211: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 211: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 212: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 212: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 213: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 213: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 214: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 214: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 215: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 215: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 216: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 216: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 217: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 217: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 218: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 218: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 219: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 219: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 220: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 220: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 221: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 221: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 222: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 222: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 223: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 223: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 224: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 224: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 225: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 225: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 226: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 226: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 227: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 227: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 228: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 228: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 229: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 229: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 230: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 230: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 231: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 231: 100% 213/213 [00:14<00:00, 14.24it/s, loss=0] Epoch: 232: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 232: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 233: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 233: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 234: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 234: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 235: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 235: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 236: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 236: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 237: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 237: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 238: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 238: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 239: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 239: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 240: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 240: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 241: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 241: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 242: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 242: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 243: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 243: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 244: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 244: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 245: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 245: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 246: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 246: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 247: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 247: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 248: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 248: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 249: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 249: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 250: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 250: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 251: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 251: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 252: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 252: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 253: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 253: 100% 213/213 [00:14<00:00, 14.32it/s, loss=0] Epoch: 254: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 254: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 255: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 255: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 256: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 256: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 257: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 257: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 258: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 258: 100% 213/213 [00:14<00:00, 14.23it/s, loss=0] Epoch: 259: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 259: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 260: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 260: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 261: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 261: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 262: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 262: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 263: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 263: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 264: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 264: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 265: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 265: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 266: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 266: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 267: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 267: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 268: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 268: 100% 213/213 [00:14<00:00, 14.28it/s, loss=0] Epoch: 269: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 269: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 270: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 270: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 271: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 271: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 272: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 272: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 273: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 273: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 274: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 274: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 275: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 275: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 276: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 276: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 277: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 277: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 278: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 278: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 279: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 279: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 280: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 280: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 281: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 281: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 282: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 282: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 283: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 283: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 284: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 284: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 285: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 285: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 286: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 286: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 287: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 287: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 288: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 288: 100% 213/213 [00:14<00:00, 14.22it/s, loss=0] Epoch: 289: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 289: 100% 213/213 [00:14<00:00, 14.26it/s, loss=0] Epoch: 290: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 290: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 291: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 291: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 292: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 292: 100% 213/213 [00:14<00:00, 14.25it/s, loss=0] Epoch: 293: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 293: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 294: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 294: 100% 213/213 [00:14<00:00, 14.29it/s, loss=0] Epoch: 295: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 295: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 296: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 296: 100% 213/213 [00:14<00:00, 14.27it/s, loss=0] Epoch: 297: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 297: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 298: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 298: 100% 213/213 [00:14<00:00, 14.30it/s, loss=0] Epoch: 299: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 299: 100% 213/213 [00:14<00:00, 14.31it/s, loss=0] Epoch: 100% 300/300 [1:14:56<00:00, 14.99s/it] INFO:root:Training is finished ... Iteration: 100% 24/24 [00:14<00:00, 1.70it/s, acc=1] Iteration: 100% 165/165 [25:41<00:00, 9.34s/it, acc=0.908] obj_pp_to_subj_pp: 11.8 cp_recursion: 57.3 pp_recursion: 62.2 subj_to_obj_proper: 91.2 prim_to_obj_proper: 86.4 prim_to_subj_proper: 99.9 LEX: 99.92666666666666 OVERALL: 90.84285714285714
seed 42: 16.6
seed 43: 16.8
seed 66: 13.0
seed 67: 18.1
seed 77: 17.0
seed 78: 17.5
seed 88: 16.2
seed 89: n/a (reported below and recomputed including, not in original draft as was not available in time (half of these seeds were run on separate run and were not complete, see below))
seed 99: 13.7
seed 100: n/a
16.1125 +/- 1.6914767955842616 (sample +/- std)
95% CI 14.940366381870223 to 17.284633618129778
after original reporting deadline, seed 89 and seed 100 of original run completed (see prior Transformer train/eval cell):
obj_pp_to_subj_pp, seed 89: 21.5%
obj_pp_to_subj_pp, seed 100: 11.8%
>>> wu_et_al_baseline_2023_3_layers_obj_pp_to_subj_pp = np.array([16.6,16.8,13.0,18.1,17.0,17.5,16.2,21.5,13.7,11.8])
wu_et_al_baseline_2023_3_layers_obj_pp_to_subj_pp.mean()
wu_et_al_baseline_2023_3_layers_obj_pp_to_subj_pp.std()
len(wu_et_al_baseline_2023_3_layers_obj_pp_to_subj_pp)
wu_et_al_baseline_2023_3_layers_obj_pp_to_subj_pp.std()/math.sqrt(10)
wu_et_al_baseline_2023_3_layers_obj_pp_to_subj_pp.mean()
(wu_et_al_baseline_2023_3_layers_obj_pp_to_subj_pp.mean() - wu_et_al_baseline_2023_3_layers_obj_pp_to_subj_pp.std()/math.sqrt(10)*1.96, wu_et_al_baseline_2023_3_layers_obj_pp_to_subj_pp.mean() + wu_et_al_baseline_2023_3_layers_obj_pp_to_subj_pp.std()/math.sqrt(10) * 1.96)
>>> 16.22
>>> 2.653224453377437
>>> 10
>>> 0.8390232416327928
>>> 16.22
>>> (14.575514446399724, 17.86448555360027)
So for the Wu et al 2023 baseline Transformer Encoder-Decoder with 3 layers we update from n=8:
16.1125 +/- 1.6914767955842616 (sample +/- std)
95% CI 14.940366381870223 to 17.284633618129778
to n=10:
16.22 +/- 2.653224453377437 (sample +/- std)
95% CI 14.575514446399724 to 17.86448555360027
Wu et al 2023 baseline Encoder-Decoder - 4 layers - not controlling for parameter count¶
ReCOGS commit used 1b6eca8ff4dca5fd2fb284a7d470998af5083beb
In [ ]:
%cd /content/
!rm -rf ReCOGS
!git clone https://github.com/frankaging/ReCOGS.git
%cd ReCOGS
!echo '{\
"architectures": [\
"Bert"\
],\
"model_type": "bert",\
"attention_probs_dropout_prob": 0.1,\
"hidden_act": "gelu",\
"hidden_dropout_prob": 0.1,\
"hidden_size": 300,\
"initializer_range": 0.02,\
"intermediate_size": 512,\
"num_attention_heads": 4,\
"num_hidden_layers": 4,\
"type_vocab_size": 2,\
"vocab_size": 762,\
"pad_token_id": 0,\
"bos_token_id": 1,\
"eos_token_id": 2,\
"unk_token_id": 3,\
"mask_token_id": 4,\
"cls_token_id": 5,\
"sum_token_id": 6,\
"nsp_token_id": 7,\
"position_embedding_type": "absolute",\
"position_embedding_init": "random"\
}' > /content/ReCOGS/model/encoder_config.json
!echo '{\
"architectures": [\
"Bert"\
],\
"model_type": "bert",\
"attention_probs_dropout_prob": 0.1,\
"hidden_act": "gelu",\
"hidden_dropout_prob": 0.1,\
"hidden_size": 300,\
"initializer_range": 0.02,\
"intermediate_size": 512,\
"num_attention_heads": 4,\
"num_hidden_layers": 4,\
"type_vocab_size": 2,\
"vocab_size": 729,\
"pad_token_id": 0,\
"bos_token_id": 1,\
"eos_token_id": 2,\
"unk_token_id": 3,\
"mask_token_id": 4,\
"decoder_start_token_id": 1,\
"position_embedding_type": "absolute",\
"position_embedding_init": "random"\
}' > /content/ReCOGS/model/decoder_config.json
seed 42: obj_pp_to_subj_pp: 23.8
seed 66: obj_pp_to_subj_pp: 20.6
seed 77: obj_pp_to_subj_pp: 23.2
seed 88: obj_pp_to_subj_pp: 16.4
seed 99: obj_pp_to_subj_pp: 12.1
19.22 +/- 4.4128902093752576 (sample mean +/- std)
95% confidence interval 15.4% to 23.1%
In [ ]:
!python run_cogs.py --model_name ende_transformer --gpu 1 --train_batch_size 128 --eval_batch_size 128 --lr 0.0001 --data_path ./recogs_positional_index --output_dir ./results_recogs_positional_index --lfs cogs --do_train --do_test --do_gen --max_seq_len 512 --output_json --epochs 300 --seeds "42;66;77;88;99" # paper's seeds, not sure how they chose them
INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 7749325 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.42 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]We strongly recommend passing in an `attention_mask` since your input_ids may be padded. See https://huggingface.co/docs/transformers/troubleshooting#incorrect-output-when-padding-tokens-arent-masked. /content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:20<00:00, 10.52it/s, loss=5.66] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:19<00:00, 11.11it/s, loss=4.46] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:19<00:00, 11.12it/s, loss=3.33] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:19<00:00, 11.11it/s, loss=2.32] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:19<00:00, 11.06it/s, loss=1.8] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:19<00:00, 11.10it/s, loss=1.43] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:19<00:00, 11.10it/s, loss=1.17] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:19<00:00, 11.07it/s, loss=1] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0.89] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0.8] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.7] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0.61] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0.53] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.46] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.39] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.32] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.27] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.22] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.18] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0.17] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.13] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.11] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0.1] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.08] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.07] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.07] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0.06] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.05] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0.06] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.05] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.04] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0.04] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0.04] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0.03] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0.03] Epoch: 35: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 35: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.03] Epoch: 36: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 36: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0.02] Epoch: 37: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 37: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.02] Epoch: 38: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 38: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.02] Epoch: 39: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 39: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.02] Epoch: 40: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 40: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0.02] Epoch: 41: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 41: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0.02] Epoch: 42: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 42: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.01] Epoch: 43: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 43: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.01] Epoch: 44: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 44: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.01] Epoch: 45: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 45: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0.02] Epoch: 46: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 46: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0.01] Epoch: 47: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 47: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0.01] Epoch: 48: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 48: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0.01] Epoch: 49: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 49: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0.01] Epoch: 50: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 50: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0.01] Epoch: 51: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 51: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.01] Epoch: 52: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 52: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0.01] Epoch: 53: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 53: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0.01] Epoch: 54: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 54: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.01] Epoch: 55: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 55: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0.01] Epoch: 56: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 56: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0.02] Epoch: 57: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 57: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.01] Epoch: 58: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 58: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0.01] Epoch: 59: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 59: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.01] Epoch: 60: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 60: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 61: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 61: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 62: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 62: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.01] Epoch: 63: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 63: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.01] Epoch: 64: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 64: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0.01] Epoch: 65: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 65: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 66: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 66: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.01] Epoch: 67: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 67: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 68: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 68: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 69: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 69: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 70: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 70: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 71: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 71: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 72: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 72: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 73: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 73: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0.01] Epoch: 74: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 74: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 75: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 75: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 76: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 76: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 77: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 77: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 78: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 78: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 79: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 79: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.01] Epoch: 80: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 80: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 81: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 81: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 82: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 82: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 83: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 83: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0.01] Epoch: 84: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 84: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.01] Epoch: 85: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 85: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 86: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 86: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 87: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 87: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 88: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 88: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 89: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 89: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 90: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 90: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 91: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 91: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 92: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 92: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 93: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 93: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 94: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 94: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 95: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 95: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.01] Epoch: 96: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 96: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 97: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 97: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 98: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 98: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 99: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 99: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 100: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 100: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 101: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 101: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 102: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 102: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 103: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 103: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 104: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 104: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 105: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 105: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 106: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 106: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 107: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 107: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 108: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 108: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 109: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 109: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 110: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 110: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 111: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 111: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 112: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 112: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 113: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 113: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 114: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 114: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 115: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 115: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 116: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 116: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 117: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 117: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 118: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 118: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 119: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 119: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 120: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 120: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 121: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 121: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 122: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 122: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 123: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 123: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.01] Epoch: 124: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 124: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 125: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 125: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 126: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 126: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 127: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 127: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 128: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 128: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 129: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 129: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 130: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 130: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 131: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 131: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 132: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 132: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 133: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 133: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 134: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 134: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 135: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 135: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 136: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 136: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 137: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 137: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 138: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 138: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 139: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 139: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.01] Epoch: 140: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 140: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 141: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 141: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 142: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 142: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 143: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 143: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 144: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 144: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 145: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 145: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 146: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 146: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 147: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 147: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 148: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 148: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 149: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 149: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 150: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 150: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 151: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 151: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 152: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 152: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 153: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 153: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 154: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 154: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 155: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 155: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 156: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 156: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 157: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 157: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 158: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 158: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 159: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 159: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 160: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 160: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 161: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 161: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 162: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 162: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 163: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 163: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 164: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 164: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 165: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 165: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 166: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 166: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 167: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 167: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 168: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 168: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 169: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 169: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 170: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 170: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 171: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 171: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 172: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 172: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 173: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 173: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 174: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 174: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 175: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 175: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 176: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 176: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 177: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 177: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 178: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 178: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 179: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 179: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 180: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 180: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 181: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 181: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 182: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 182: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 183: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 183: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 184: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 184: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 185: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 185: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 186: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 186: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 187: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 187: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 188: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 188: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 189: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 189: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 190: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 190: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 191: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 191: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 192: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 192: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 193: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 193: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 194: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 194: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 195: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 195: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 196: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 196: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 197: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 197: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 198: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 198: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 199: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 199: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 200: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 200: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 201: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 201: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 202: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 202: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 203: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 203: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 204: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 204: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 205: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 205: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 206: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 206: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 207: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 207: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 208: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 208: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 209: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 209: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 210: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 210: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 211: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 211: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 212: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 212: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 213: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 213: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 214: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 214: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 215: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 215: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 216: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 216: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 217: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 217: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 218: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 218: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 219: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 219: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 220: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 220: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 221: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 221: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 222: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 222: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 223: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 223: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 224: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 224: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 225: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 225: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 226: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 226: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 227: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 227: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 228: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 228: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 229: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 229: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 230: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 230: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 231: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 231: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 232: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 232: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 233: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 233: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 234: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 234: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 235: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 235: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 236: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 236: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 237: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 237: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 238: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 238: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 239: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 239: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 240: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 240: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 241: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 241: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 242: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 242: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 243: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 243: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 244: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 244: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 245: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 245: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 246: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 246: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 247: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 247: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 248: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 248: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 249: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 249: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 250: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 250: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 251: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 251: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 252: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 252: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 253: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 253: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 254: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 254: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 255: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 255: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 256: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 256: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 257: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 257: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 258: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 258: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 259: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 259: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 260: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 260: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 261: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 261: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 262: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 262: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 263: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 263: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 264: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 264: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 265: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 265: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 266: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 266: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 267: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 267: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 268: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 268: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 269: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 269: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 270: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 270: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 271: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 271: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 272: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 272: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 273: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 273: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 274: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 274: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 275: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 275: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 276: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 276: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 277: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 277: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 278: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 278: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 279: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 279: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 280: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 280: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 281: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 281: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 282: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 282: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 283: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 283: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 284: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 284: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 285: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 285: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 286: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 286: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 287: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 287: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 288: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 288: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 289: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 289: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 290: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 290: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 291: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 291: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 292: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 292: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 293: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 293: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 294: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 294: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 295: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 295: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 296: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 296: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 297: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 297: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 298: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 298: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 299: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 299: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 100% 300/300 [1:36:42<00:00, 19.34s/it] INFO:root:Training is finished ... Iteration: 100% 24/24 [00:17<00:00, 1.39it/s, acc=1] Iteration: 100% 165/165 [21:16<00:00, 7.73s/it, acc=0.904] obj_pp_to_subj_pp: 23.8 cp_recursion: 58.0 pp_recursion: 26.0 subj_to_obj_proper: 96.7 prim_to_obj_proper: 94.2 prim_to_subj_proper: 100.0 LEX: 99.94666666666666 OVERALL: 90.37619047619047 INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 7749325 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.66 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:19<00:00, 11.07it/s, loss=5.66] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:19<00:00, 11.08it/s, loss=4.49] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:19<00:00, 11.06it/s, loss=3.42] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:19<00:00, 11.08it/s, loss=2.39] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:19<00:00, 11.07it/s, loss=1.86] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:19<00:00, 11.07it/s, loss=1.5] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:19<00:00, 11.08it/s, loss=1.23] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:19<00:00, 11.08it/s, loss=1.05] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.92] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.8] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.71] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0.62] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.52] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.45] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.37] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0.31] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0.26] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.2] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.18] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.16] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.12] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0.11] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.09] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.07] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0.07] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.06] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.07] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0.05] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0.04] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0.04] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.05] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0.03] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.02] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.02] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.02] Epoch: 35: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 35: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.04] Epoch: 36: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 36: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.02] Epoch: 37: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 37: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.02] Epoch: 38: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 38: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.02] Epoch: 39: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 39: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0.02] Epoch: 40: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 40: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.02] Epoch: 41: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 41: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0.01] Epoch: 42: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 42: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.01] Epoch: 43: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 43: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.02] Epoch: 44: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 44: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.02] Epoch: 45: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 45: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.01] Epoch: 46: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 46: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.01] Epoch: 47: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 47: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.01] Epoch: 48: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 48: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.01] Epoch: 49: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 49: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.01] Epoch: 50: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 50: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0.01] Epoch: 51: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 51: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.01] Epoch: 52: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 52: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.01] Epoch: 53: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 53: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0.01] Epoch: 54: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 54: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0.01] Epoch: 55: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 55: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.01] Epoch: 56: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 56: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 57: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 57: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0.01] Epoch: 58: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 58: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.01] Epoch: 59: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 59: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0.01] Epoch: 60: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 60: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.01] Epoch: 61: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 61: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.01] Epoch: 62: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 62: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0.01] Epoch: 63: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 63: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.01] Epoch: 64: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 64: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 65: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 65: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.01] Epoch: 66: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 66: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 67: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 67: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 68: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 68: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 69: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 69: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 70: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 70: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 71: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 71: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0.01] Epoch: 72: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 72: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0.01] Epoch: 73: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 73: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 74: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 74: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 75: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 75: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 76: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 76: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 77: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 77: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 78: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 78: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 79: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 79: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 80: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 80: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 81: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 81: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 82: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 82: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.01] Epoch: 83: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 83: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.01] Epoch: 84: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 84: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 85: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 85: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 86: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 86: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 87: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 87: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 88: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 88: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 89: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 89: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 90: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 90: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 91: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 91: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 92: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 92: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 93: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 93: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 94: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 94: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 95: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 95: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 96: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 96: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 97: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 97: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 98: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 98: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 99: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 99: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 100: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 100: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 101: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 101: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 102: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 102: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 103: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 103: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 104: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 104: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 105: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 105: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 106: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 106: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 107: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 107: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 108: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 108: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 109: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 109: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 110: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 110: 100% 213/213 [00:19<00:00, 11.00it/s, loss=0] Epoch: 111: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 111: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0] Epoch: 112: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 112: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 113: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 113: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0] Epoch: 114: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 114: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 115: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 115: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 116: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 116: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 117: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 117: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 118: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 118: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 119: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 119: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 120: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 120: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 121: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 121: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 122: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 122: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 123: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 123: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 124: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 124: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 125: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 125: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 126: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 126: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 127: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 127: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 128: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 128: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 129: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 129: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 130: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 130: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 131: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 131: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 132: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 132: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 133: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 133: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 134: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 134: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 135: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 135: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 136: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 136: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 137: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 137: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 138: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 138: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 139: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 139: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 140: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 140: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 141: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 141: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 142: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 142: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 143: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 143: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 144: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 144: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 145: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 145: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 146: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 146: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 147: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 147: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 148: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 148: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 149: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 149: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 150: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 150: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 151: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 151: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 152: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 152: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 153: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 153: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 154: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 154: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 155: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 155: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 156: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 156: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 157: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 157: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 158: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 158: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 159: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 159: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 160: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 160: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 161: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 161: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 162: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 162: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 163: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 163: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 164: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 164: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 165: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 165: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 166: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 166: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 167: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 167: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 168: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 168: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 169: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 169: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 170: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 170: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 171: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 171: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 172: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 172: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 173: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 173: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 174: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 174: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 175: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 175: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 176: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 176: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 177: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 177: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 178: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 178: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 179: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 179: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 180: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 180: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 181: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 181: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 182: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 182: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 183: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 183: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 184: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 184: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 185: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 185: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 186: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 186: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 187: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 187: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 188: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 188: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 189: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 189: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 190: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 190: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 191: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 191: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 192: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 192: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 193: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 193: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 194: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 194: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 195: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 195: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 196: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 196: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 197: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 197: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 198: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 198: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 199: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 199: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 200: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 200: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 201: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 201: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 202: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 202: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 203: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 203: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 204: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 204: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 205: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 205: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 206: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 206: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 207: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 207: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 208: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 208: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 209: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 209: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 210: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 210: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 211: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 211: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 212: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 212: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 213: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 213: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 214: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 214: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 215: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 215: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 216: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 216: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 217: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 217: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 218: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 218: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 219: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 219: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 220: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 220: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 221: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 221: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 222: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 222: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 223: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 223: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 224: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 224: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 225: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 225: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 226: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 226: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 227: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 227: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 228: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 228: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 229: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 229: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 230: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 230: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 231: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 231: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 232: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 232: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 233: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 233: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 234: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 234: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 235: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 235: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 236: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 236: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 237: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 237: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 238: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 238: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 239: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 239: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 240: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 240: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 241: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 241: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 242: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 242: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 243: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 243: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 244: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 244: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 245: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 245: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 246: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 246: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 247: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 247: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 248: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 248: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 249: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 249: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 250: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 250: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 251: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 251: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 252: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 252: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 253: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 253: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 254: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 254: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 255: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 255: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 256: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 256: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 257: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 257: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 258: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 258: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 259: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 259: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 260: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 260: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 261: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 261: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 262: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 262: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 263: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 263: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 264: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 264: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 265: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 265: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 266: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 266: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 267: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 267: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 268: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 268: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0] Epoch: 269: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 269: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 270: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 270: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 271: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 271: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 272: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 272: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 273: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 273: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 274: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 274: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 275: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 275: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 276: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 276: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 277: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 277: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 278: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 278: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 279: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 279: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 280: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 280: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 281: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 281: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 282: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 282: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 283: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 283: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 284: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 284: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0] Epoch: 285: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 285: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 286: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 286: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 287: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 287: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 288: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 288: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 289: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 289: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 290: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 290: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 291: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 291: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 292: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 292: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 293: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 293: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 294: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 294: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 295: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 295: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 296: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 296: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 297: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 297: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 298: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 298: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 299: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 299: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 100% 300/300 [1:36:42<00:00, 19.34s/it] INFO:root:Training is finished ... Iteration: 100% 24/24 [00:17<00:00, 1.36it/s, acc=1] Iteration: 100% 165/165 [37:34<00:00, 13.67s/it, acc=0.91] obj_pp_to_subj_pp: 20.6 cp_recursion: 63.1 pp_recursion: 46.0 subj_to_obj_proper: 91.7 prim_to_obj_proper: 91.4 prim_to_subj_proper: 99.9 LEX: 99.94666666666666 OVERALL: 91.04285714285714 INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 7749325 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.77 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:19<00:00, 11.03it/s, loss=5.7] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:19<00:00, 11.07it/s, loss=4.5] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:19<00:00, 11.08it/s, loss=3.43] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:19<00:00, 11.08it/s, loss=2.43] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:19<00:00, 11.08it/s, loss=1.88] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:19<00:00, 11.09it/s, loss=1.54] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:19<00:00, 11.08it/s, loss=1.28] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:19<00:00, 11.08it/s, loss=1.11] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.98] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.88] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0.78] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.69] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0.59] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0.51] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.43] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0.36] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:19<00:00, 11.00it/s, loss=0.29] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0.24] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0.21] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0.16] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0.14] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0.13] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0.11] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0.09] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0.08] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.07] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0.07] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0.06] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.05] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.05] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0.05] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0.04] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.03] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.04] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.03] Epoch: 35: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 35: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.03] Epoch: 36: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 36: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.03] Epoch: 37: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 37: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.03] Epoch: 38: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 38: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.03] Epoch: 39: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 39: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0.02] Epoch: 40: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 40: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.03] Epoch: 41: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 41: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.03] Epoch: 42: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 42: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.02] Epoch: 43: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 43: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.02] Epoch: 44: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 44: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.02] Epoch: 45: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 45: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0.02] Epoch: 46: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 46: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.01] Epoch: 47: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 47: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.01] Epoch: 48: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 48: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0.01] Epoch: 49: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 49: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.01] Epoch: 50: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 50: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.01] Epoch: 51: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 51: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0.01] Epoch: 52: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 52: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.01] Epoch: 53: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 53: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.01] Epoch: 54: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 54: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.01] Epoch: 55: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 55: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.01] Epoch: 56: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 56: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0.01] Epoch: 57: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 57: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0.01] Epoch: 58: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 58: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0.01] Epoch: 59: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 59: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.01] Epoch: 60: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 60: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.01] Epoch: 61: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 61: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0.01] Epoch: 62: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 62: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.01] Epoch: 63: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 63: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.01] Epoch: 64: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 64: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.01] Epoch: 65: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 65: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.01] Epoch: 66: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 66: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0.01] Epoch: 67: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 67: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.01] Epoch: 68: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 68: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0.01] Epoch: 69: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 69: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.01] Epoch: 70: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 70: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 71: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 71: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.01] Epoch: 72: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 72: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 73: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 73: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 74: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 74: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0.01] Epoch: 75: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 75: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.01] Epoch: 76: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 76: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.01] Epoch: 77: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 77: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 78: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 78: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 79: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 79: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 80: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 80: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 81: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 81: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.01] Epoch: 82: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 82: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 83: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 83: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 84: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 84: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.01] Epoch: 85: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 85: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.01] Epoch: 86: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 86: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 87: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 87: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0.01] Epoch: 88: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 88: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 89: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 89: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 90: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 90: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 91: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 91: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 92: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 92: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 93: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 93: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 94: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 94: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 95: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 95: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0.01] Epoch: 96: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 96: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 97: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 97: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 98: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 98: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0.01] Epoch: 99: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 99: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 100: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 100: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 101: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 101: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 102: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 102: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 103: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 103: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0.01] Epoch: 104: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 104: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 105: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 105: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 106: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 106: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0] Epoch: 107: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 107: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 108: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 108: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 109: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 109: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 110: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 110: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0] Epoch: 111: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 111: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0] Epoch: 112: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 112: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 113: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 113: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0] Epoch: 114: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 114: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 115: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 115: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 116: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 116: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 117: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 117: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 118: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 118: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 119: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 119: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 120: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 120: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 121: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 121: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 122: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 122: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 123: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 123: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 124: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 124: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 125: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 125: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 126: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 126: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 127: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 127: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 128: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 128: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 129: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 129: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 130: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 130: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 131: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 131: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 132: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 132: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 133: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 133: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 134: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 134: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 135: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 135: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 136: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 136: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 137: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 137: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 138: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 138: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 139: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 139: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 140: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 140: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 141: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 141: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 142: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 142: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 143: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 143: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0] Epoch: 144: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 144: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0.01] Epoch: 145: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 145: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 146: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 146: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 147: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 147: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 148: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 148: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 149: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 149: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 150: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 150: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 151: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 151: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 152: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 152: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 153: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 153: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 154: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 154: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 155: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 155: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 156: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 156: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 157: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 157: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 158: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 158: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 159: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 159: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 160: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 160: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 161: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 161: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 162: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 162: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 163: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 163: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 164: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 164: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 165: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 165: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 166: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 166: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 167: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 167: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 168: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 168: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0] Epoch: 169: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 169: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 170: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 170: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 171: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 171: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 172: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 172: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 173: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 173: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 174: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 174: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 175: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 175: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 176: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 176: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 177: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 177: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 178: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 178: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 179: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 179: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0] Epoch: 180: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 180: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0] Epoch: 181: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 181: 100% 213/213 [00:19<00:00, 11.12it/s, loss=0] Epoch: 182: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 182: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 183: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 183: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 184: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 184: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 185: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 185: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 186: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 186: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 187: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 187: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 188: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 188: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 189: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 189: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 190: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 190: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 191: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 191: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 192: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 192: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 193: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 193: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 194: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 194: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 195: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 195: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 196: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 196: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 197: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 197: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 198: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 198: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 199: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 199: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 200: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 200: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 201: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 201: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 202: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 202: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 203: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 203: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 204: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 204: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 205: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 205: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 206: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 206: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 207: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 207: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 208: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 208: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 209: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 209: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 210: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 210: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 211: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 211: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 212: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 212: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 213: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 213: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 214: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 214: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 215: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 215: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 216: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 216: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 217: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 217: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 218: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 218: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 219: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 219: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 220: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 220: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 221: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 221: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 222: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 222: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 223: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 223: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 224: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 224: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 225: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 225: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 226: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 226: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 227: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 227: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 228: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 228: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 229: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 229: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 230: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 230: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 231: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 231: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 232: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 232: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 233: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 233: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 234: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 234: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 235: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 235: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 236: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 236: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 237: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 237: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 238: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 238: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 239: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 239: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 240: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 240: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 241: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 241: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 242: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 242: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 243: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 243: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 244: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 244: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 245: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 245: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 246: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 246: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 247: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 247: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 248: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 248: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 249: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 249: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 250: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 250: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 251: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 251: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 252: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 252: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 253: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 253: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 254: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 254: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 255: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 255: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 256: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 256: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 257: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 257: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0] Epoch: 258: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 258: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0] Epoch: 259: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 259: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 260: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 260: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0] Epoch: 261: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 261: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0] Epoch: 262: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 262: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 263: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 263: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 264: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 264: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 265: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 265: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 266: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 266: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 267: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 267: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 268: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 268: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0] Epoch: 269: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 269: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 270: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 270: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0] Epoch: 271: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 271: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 272: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 272: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 273: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 273: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 274: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 274: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 275: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 275: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 276: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 276: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 277: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 277: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 278: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 278: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0] Epoch: 279: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 279: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0] Epoch: 280: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 280: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 281: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 281: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 282: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 282: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0] Epoch: 283: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 283: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 284: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 284: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0] Epoch: 285: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 285: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0] Epoch: 286: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 286: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 287: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 287: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 288: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 288: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 289: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 289: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0] Epoch: 290: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 290: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 291: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 291: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 292: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 292: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0] Epoch: 293: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 293: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 294: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 294: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 295: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 295: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 296: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 296: 100% 213/213 [00:19<00:00, 11.08it/s, loss=0] Epoch: 297: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 297: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 298: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 298: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 299: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 299: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 100% 300/300 [1:36:40<00:00, 19.34s/it] INFO:root:Training is finished ... Iteration: 100% 24/24 [00:17<00:00, 1.37it/s, acc=1] Iteration: 100% 165/165 [33:39<00:00, 12.24s/it, acc=0.903] obj_pp_to_subj_pp: 23.2 cp_recursion: 62.5 pp_recursion: 30.9 subj_to_obj_proper: 89.0 prim_to_obj_proper: 92.0 prim_to_subj_proper: 100.0 LEX: 99.94666666666666 OVERALL: 90.32380952380953 INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 7749325 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.88 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:19<00:00, 11.09it/s, loss=5.64] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:19<00:00, 11.15it/s, loss=4.46] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:19<00:00, 11.15it/s, loss=3.35] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:19<00:00, 11.15it/s, loss=2.28] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:19<00:00, 11.18it/s, loss=1.71] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:19<00:00, 11.15it/s, loss=1.34] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:19<00:00, 11.18it/s, loss=1.08] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0.9] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0.76] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:19<00:00, 11.18it/s, loss=0.65] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:19<00:00, 11.18it/s, loss=0.55] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0.45] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0.35] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0.28] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0.22] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0.16] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:19<00:00, 11.18it/s, loss=0.13] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0.1] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0.08] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0.06] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0.05] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0.04] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0.04] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0.03] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0.02] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0.02] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0.02] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:19<00:00, 11.19it/s, loss=0.01] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0.01] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0.01] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0.01] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0.01] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0.01] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 35: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 35: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0.01] Epoch: 36: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 36: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0] Epoch: 37: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 37: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 38: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 38: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 39: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 39: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 40: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 40: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0.01] Epoch: 41: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 41: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 42: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 42: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 43: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 43: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0] Epoch: 44: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 44: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 45: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 45: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0] Epoch: 46: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 46: 100% 213/213 [00:19<00:00, 11.18it/s, loss=0] Epoch: 47: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 47: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 48: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 48: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0] Epoch: 49: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 49: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0] Epoch: 50: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 50: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0] Epoch: 51: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 51: 100% 213/213 [00:19<00:00, 11.18it/s, loss=0] Epoch: 52: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 52: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 53: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 53: 100% 213/213 [00:19<00:00, 11.12it/s, loss=0] Epoch: 54: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 54: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0] Epoch: 55: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 55: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 56: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 56: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 57: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 57: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 58: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 58: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0] Epoch: 59: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 59: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 60: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 60: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 61: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 61: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0] Epoch: 62: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 62: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 63: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 63: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 64: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 64: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0] Epoch: 65: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 65: 100% 213/213 [00:19<00:00, 11.18it/s, loss=0] Epoch: 66: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 66: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0] Epoch: 67: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 67: 100% 213/213 [00:19<00:00, 11.19it/s, loss=0] Epoch: 68: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 68: 100% 213/213 [00:19<00:00, 11.18it/s, loss=0] Epoch: 69: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 69: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 70: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 70: 100% 213/213 [00:19<00:00, 11.18it/s, loss=0] Epoch: 71: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 71: 100% 213/213 [00:19<00:00, 11.18it/s, loss=0] Epoch: 72: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 72: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0] Epoch: 73: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 73: 100% 213/213 [00:19<00:00, 11.18it/s, loss=0] Epoch: 74: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 74: 100% 213/213 [00:19<00:00, 11.18it/s, loss=0] Epoch: 75: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 75: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0] Epoch: 76: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 76: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0] Epoch: 77: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 77: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 78: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 78: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0] Epoch: 79: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 79: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 80: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 80: 100% 213/213 [00:19<00:00, 11.18it/s, loss=0] Epoch: 81: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 81: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0] Epoch: 82: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 82: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 83: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 83: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0] Epoch: 84: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 84: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0] Epoch: 85: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 85: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 86: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 86: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0] Epoch: 87: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 87: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 88: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 88: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 89: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 89: 100% 213/213 [00:19<00:00, 11.18it/s, loss=0] Epoch: 90: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 90: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0] Epoch: 91: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 91: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0] Epoch: 92: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 92: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 93: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 93: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 94: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 94: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0] Epoch: 95: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 95: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 96: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 96: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0] Epoch: 97: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 97: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0] Epoch: 98: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 98: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 99: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 99: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 100: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 100: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 101: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 101: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 102: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 102: 100% 213/213 [00:19<00:00, 11.18it/s, loss=0] Epoch: 103: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 103: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 104: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 104: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0] Epoch: 105: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 105: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 106: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 106: 100% 213/213 [00:19<00:00, 11.18it/s, loss=0] Epoch: 107: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 107: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0] Epoch: 108: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 108: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 109: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 109: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 110: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 110: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0] Epoch: 111: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 111: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 112: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 112: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0] Epoch: 113: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 113: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0] Epoch: 114: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 114: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 115: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 115: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 116: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 116: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 117: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 117: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 118: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 118: 100% 213/213 [00:19<00:00, 11.18it/s, loss=0] Epoch: 119: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 119: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 120: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 120: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 121: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 121: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0] Epoch: 122: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 122: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 123: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 123: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0] Epoch: 124: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 124: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 125: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 125: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 126: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 126: 100% 213/213 [00:19<00:00, 11.18it/s, loss=0] Epoch: 127: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 127: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 128: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 128: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 129: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 129: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 130: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 130: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 131: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 131: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 132: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 132: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 133: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 133: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 134: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 134: 100% 213/213 [00:19<00:00, 11.18it/s, loss=0] Epoch: 135: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 135: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 136: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 136: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 137: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 137: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 138: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 138: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 139: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 139: 100% 213/213 [00:19<00:00, 11.13it/s, loss=0] Epoch: 140: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 140: 100% 213/213 [00:19<00:00, 11.13it/s, loss=0] Epoch: 141: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 141: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 142: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 142: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 143: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 143: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 144: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 144: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 145: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 145: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 146: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 146: 100% 213/213 [00:19<00:00, 11.13it/s, loss=0] Epoch: 147: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 147: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 148: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 148: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 149: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 149: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 150: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 150: 100% 213/213 [00:19<00:00, 11.13it/s, loss=0] Epoch: 151: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 151: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 152: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 152: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 153: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 153: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 154: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 154: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0] Epoch: 155: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 155: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0] Epoch: 156: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 156: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 157: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 157: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 158: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 158: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 159: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 159: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 160: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 160: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 161: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 161: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 162: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 162: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 163: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 163: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 164: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 164: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 165: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 165: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 166: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 166: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 167: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 167: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 168: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 168: 100% 213/213 [00:19<00:00, 11.13it/s, loss=0] Epoch: 169: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 169: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 170: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 170: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 171: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 171: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 172: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 172: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 173: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 173: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0] Epoch: 174: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 174: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0] Epoch: 175: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 175: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0] Epoch: 176: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 176: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 177: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 177: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 178: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 178: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0] Epoch: 179: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 179: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 180: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 180: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 181: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 181: 100% 213/213 [00:19<00:00, 11.13it/s, loss=0] Epoch: 182: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 182: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 183: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 183: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 184: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 184: 100% 213/213 [00:19<00:00, 11.13it/s, loss=0] Epoch: 185: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 185: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 186: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 186: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 187: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 187: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 188: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 188: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 189: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 189: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 190: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 190: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 191: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 191: 100% 213/213 [00:19<00:00, 11.12it/s, loss=0] Epoch: 192: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 192: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 193: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 193: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 194: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 194: 100% 213/213 [00:19<00:00, 11.12it/s, loss=0] Epoch: 195: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 195: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 196: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 196: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 197: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 197: 100% 213/213 [00:19<00:00, 11.13it/s, loss=0] Epoch: 198: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 198: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 199: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 199: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 200: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 200: 100% 213/213 [00:19<00:00, 11.13it/s, loss=0] Epoch: 201: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 201: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 202: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 202: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 203: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 203: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 204: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 204: 100% 213/213 [00:19<00:00, 11.13it/s, loss=0] Epoch: 205: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 205: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 206: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 206: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 207: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 207: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 208: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 208: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0] Epoch: 209: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 209: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 210: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 210: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 211: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 211: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 212: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 212: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 213: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 213: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 214: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 214: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 215: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 215: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 216: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 216: 100% 213/213 [00:19<00:00, 11.13it/s, loss=0] Epoch: 217: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 217: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 218: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 218: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 219: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 219: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 220: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 220: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 221: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 221: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 222: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 222: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 223: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 223: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 224: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 224: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 225: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 225: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 226: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 226: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 227: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 227: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 228: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 228: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 229: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 229: 100% 213/213 [00:19<00:00, 11.13it/s, loss=0] Epoch: 230: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 230: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 231: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 231: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 232: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 232: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 233: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 233: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 234: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 234: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 235: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 235: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 236: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 236: 100% 213/213 [00:19<00:00, 11.13it/s, loss=0] Epoch: 237: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 237: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 238: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 238: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 239: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 239: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 240: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 240: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 241: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 241: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 242: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 242: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 243: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 243: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 244: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 244: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 245: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 245: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 246: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 246: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 247: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 247: 100% 213/213 [00:19<00:00, 11.13it/s, loss=0] Epoch: 248: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 248: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0] Epoch: 249: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 249: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 250: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 250: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 251: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 251: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 252: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 252: 100% 213/213 [00:19<00:00, 11.13it/s, loss=0] Epoch: 253: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 253: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 254: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 254: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 255: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 255: 100% 213/213 [00:19<00:00, 11.12it/s, loss=0] Epoch: 256: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 256: 100% 213/213 [00:19<00:00, 11.13it/s, loss=0] Epoch: 257: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 257: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 258: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 258: 100% 213/213 [00:19<00:00, 11.13it/s, loss=0] Epoch: 259: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 259: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 260: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 260: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 261: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 261: 100% 213/213 [00:19<00:00, 11.13it/s, loss=0] Epoch: 262: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 262: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 263: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 263: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 264: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 264: 100% 213/213 [00:19<00:00, 11.13it/s, loss=0] Epoch: 265: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 265: 100% 213/213 [00:19<00:00, 11.17it/s, loss=0] Epoch: 266: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 266: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 267: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 267: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 268: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 268: 100% 213/213 [00:19<00:00, 11.16it/s, loss=0] Epoch: 269: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 269: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 270: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 270: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 271: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 271: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 272: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 272: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 273: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 273: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 274: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 274: 100% 213/213 [00:19<00:00, 11.13it/s, loss=0] Epoch: 275: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 275: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 276: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 276: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 277: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 277: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 278: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 278: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 279: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 279: 100% 213/213 [00:19<00:00, 11.13it/s, loss=0] Epoch: 280: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 280: 100% 213/213 [00:19<00:00, 11.12it/s, loss=0] Epoch: 281: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 281: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 282: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 282: 100% 213/213 [00:19<00:00, 11.13it/s, loss=0] Epoch: 283: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 283: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 284: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 284: 100% 213/213 [00:19<00:00, 11.09it/s, loss=0] Epoch: 285: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 285: 100% 213/213 [00:19<00:00, 11.13it/s, loss=0] Epoch: 286: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 286: 100% 213/213 [00:19<00:00, 11.10it/s, loss=0] Epoch: 287: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 287: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 288: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 288: 100% 213/213 [00:19<00:00, 11.13it/s, loss=0] Epoch: 289: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 289: 100% 213/213 [00:19<00:00, 11.13it/s, loss=0] Epoch: 290: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 290: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0] Epoch: 291: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 291: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 292: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 292: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 293: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 293: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 294: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 294: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 295: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 295: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 296: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 296: 100% 213/213 [00:19<00:00, 11.11it/s, loss=0] Epoch: 297: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 297: 100% 213/213 [00:19<00:00, 11.15it/s, loss=0] Epoch: 298: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 298: 100% 213/213 [00:19<00:00, 11.14it/s, loss=0] Epoch: 299: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 299: 100% 213/213 [00:19<00:00, 11.12it/s, loss=0] Epoch: 100% 300/300 [1:36:00<00:00, 19.20s/it] INFO:root:Training is finished ... Iteration: 100% 24/24 [00:17<00:00, 1.35it/s, acc=1] Iteration: 100% 165/165 [35:13<00:00, 12.81s/it, acc=0.858] obj_pp_to_subj_pp: 16.4 cp_recursion: 58.6 pp_recursion: 29.2 subj_to_obj_proper: 93.2 prim_to_obj_proper: 94.8 prim_to_subj_proper: 100.0 LEX: 93.94666666666667 OVERALL: 85.78095238095239 INFO:root:Baselining the Transformer Encoder-Decoder Model INFO:root:__Number CUDA Devices: 1 INFO:root:Number of model params: 7749325 INFO:root:OUTPUT DIR: ./results_recogs_positional_index/cogs_pipeline.model.ende_transformer.lf.cogs.glove.False.seed.99 Epoch: 0: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 0: 100% 213/213 [00:19<00:00, 10.96it/s, loss=5.63] Epoch: 1: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 1: 100% 213/213 [00:19<00:00, 11.02it/s, loss=4.45] Epoch: 2: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 2: 100% 213/213 [00:19<00:00, 11.03it/s, loss=3.28] Epoch: 3: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 3: 100% 213/213 [00:19<00:00, 11.04it/s, loss=2.3] Epoch: 4: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 4: 100% 213/213 [00:19<00:00, 11.01it/s, loss=1.78] Epoch: 5: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 5: 100% 213/213 [00:19<00:00, 11.05it/s, loss=1.41] Epoch: 6: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 6: 100% 213/213 [00:19<00:00, 11.04it/s, loss=1.15] Epoch: 7: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 7: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0.99] Epoch: 8: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 8: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0.86] Epoch: 9: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 9: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0.74] Epoch: 10: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 10: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0.65] Epoch: 11: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 11: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0.55] Epoch: 12: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 12: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0.47] Epoch: 13: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 13: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0.4] Epoch: 14: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 14: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0.33] Epoch: 15: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 15: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0.25] Epoch: 16: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 16: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0.21] Epoch: 17: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 17: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0.16] Epoch: 18: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 18: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0.12] Epoch: 19: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 19: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0.11] Epoch: 20: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 20: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0.08] Epoch: 21: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 21: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0.07] Epoch: 22: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 22: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0.05] Epoch: 23: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 23: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0.04] Epoch: 24: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 24: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0.03] Epoch: 25: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 25: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0.03] Epoch: 26: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 26: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0.03] Epoch: 27: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 27: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0.02] Epoch: 28: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 28: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0.02] Epoch: 29: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 29: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0.02] Epoch: 30: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 30: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0.02] Epoch: 31: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 31: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0.01] Epoch: 32: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 32: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0.01] Epoch: 33: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 33: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0.01] Epoch: 34: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 34: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0.01] Epoch: 35: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 35: 100% 213/213 [00:19<00:00, 11.01it/s, loss=0.01] Epoch: 36: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 36: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0.01] Epoch: 37: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 37: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0.01] Epoch: 38: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 38: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0.01] Epoch: 39: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 39: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0.01] Epoch: 40: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 40: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 41: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 41: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 42: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 42: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0] Epoch: 43: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 43: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 44: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 44: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 45: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 45: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 46: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 46: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0.01] Epoch: 47: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 47: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 48: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 48: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 49: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 49: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 50: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 50: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 51: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 51: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 52: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 52: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 53: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 53: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 54: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 54: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 55: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 55: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 56: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 56: 100% 213/213 [00:19<00:00, 11.07it/s, loss=0] Epoch: 57: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 57: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 58: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 58: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 59: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 59: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 60: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 60: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 61: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 61: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 62: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 62: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 63: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 63: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 64: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 64: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 65: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 65: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 66: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 66: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 67: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 67: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 68: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 68: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 69: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 69: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 70: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 70: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 71: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 71: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 72: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 72: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 73: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 73: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 74: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 74: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 75: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 75: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 76: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 76: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 77: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 77: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 78: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 78: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 79: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 79: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 80: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 80: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 81: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 81: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 82: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 82: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 83: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 83: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 84: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 84: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 85: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 85: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 86: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 86: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 87: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 87: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 88: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 88: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 89: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 89: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 90: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 90: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 91: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 91: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 92: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 92: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 93: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 93: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 94: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 94: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 95: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 95: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 96: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 96: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 97: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 97: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 98: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 98: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 99: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 99: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0] Epoch: 100: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 100: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 101: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 101: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 102: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 102: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 103: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 103: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 104: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 104: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 105: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 105: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 106: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 106: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 107: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 107: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 108: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 108: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 109: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 109: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 110: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 110: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 111: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 111: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 112: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 112: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 113: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 113: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 114: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 114: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 115: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 115: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 116: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 116: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 117: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 117: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 118: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 118: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 119: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 119: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 120: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 120: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 121: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 121: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 122: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 122: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 123: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 123: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 124: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 124: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 125: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 125: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 126: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 126: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 127: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 127: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 128: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 128: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 129: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 129: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 130: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 130: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 131: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 131: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 132: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 132: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 133: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 133: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0] Epoch: 134: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 134: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 135: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 135: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 136: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 136: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 137: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 137: 100% 213/213 [00:19<00:00, 11.01it/s, loss=0] Epoch: 138: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 138: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0] Epoch: 139: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 139: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0] Epoch: 140: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 140: 100% 213/213 [00:19<00:00, 10.98it/s, loss=0] Epoch: 141: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 141: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 142: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 142: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0] Epoch: 143: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 143: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0] Epoch: 144: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 144: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 145: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 145: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 146: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 146: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 147: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 147: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0] Epoch: 148: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 148: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 149: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 149: 100% 213/213 [00:19<00:00, 11.01it/s, loss=0] Epoch: 150: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 150: 100% 213/213 [00:19<00:00, 11.01it/s, loss=0] Epoch: 151: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 151: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 152: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 152: 100% 213/213 [00:19<00:00, 11.01it/s, loss=0] Epoch: 153: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 153: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 154: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 154: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 155: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 155: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 156: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 156: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 157: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 157: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 158: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 158: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 159: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 159: 100% 213/213 [00:19<00:00, 11.01it/s, loss=0] Epoch: 160: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 160: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 161: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 161: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 162: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 162: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 163: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 163: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 164: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 164: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0] Epoch: 165: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 165: 100% 213/213 [00:19<00:00, 11.00it/s, loss=0] Epoch: 166: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 166: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 167: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 167: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 168: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 168: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0] Epoch: 169: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 169: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 170: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 170: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 171: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 171: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0] Epoch: 172: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 172: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 173: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 173: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 174: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 174: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 175: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 175: 100% 213/213 [00:19<00:00, 11.00it/s, loss=0] Epoch: 176: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 176: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 177: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 177: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 178: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 178: 100% 213/213 [00:19<00:00, 11.00it/s, loss=0] Epoch: 179: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 179: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0] Epoch: 180: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 180: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 181: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 181: 100% 213/213 [00:19<00:00, 11.01it/s, loss=0] Epoch: 182: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 182: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 183: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 183: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0] Epoch: 184: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 184: 100% 213/213 [00:19<00:00, 11.00it/s, loss=0] Epoch: 185: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 185: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0] Epoch: 186: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 186: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 187: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 187: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 188: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 188: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 189: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 189: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 190: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 190: 100% 213/213 [00:19<00:00, 11.01it/s, loss=0] Epoch: 191: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 191: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 192: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 192: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 193: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 193: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 194: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 194: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 195: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 195: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 196: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 196: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 197: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 197: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 198: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 198: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 199: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 199: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 200: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 200: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0] Epoch: 201: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 201: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 202: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 202: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 203: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 203: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 204: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 204: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 205: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 205: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 206: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 206: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0] Epoch: 207: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 207: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 208: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 208: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 209: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 209: 100% 213/213 [00:19<00:00, 11.01it/s, loss=0] Epoch: 210: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 210: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 211: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 211: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 212: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 212: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 213: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 213: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 214: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 214: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 215: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 215: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 216: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 216: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 217: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 217: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 218: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 218: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 219: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 219: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 220: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 220: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 221: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 221: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 222: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 222: 100% 213/213 [00:19<00:00, 11.01it/s, loss=0] Epoch: 223: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 223: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 224: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 224: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 225: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 225: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 226: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 226: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 227: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 227: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 228: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 228: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0] Epoch: 229: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 229: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 230: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 230: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 231: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 231: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 232: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 232: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 233: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 233: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 234: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 234: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 235: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 235: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0] Epoch: 236: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 236: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 237: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 237: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 238: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 238: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0] Epoch: 239: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 239: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 240: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 240: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 241: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 241: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 242: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 242: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 243: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 243: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 244: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 244: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 245: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 245: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 246: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 246: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 247: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 247: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0] Epoch: 248: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 248: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 249: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 249: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 250: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 250: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 251: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 251: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 252: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 252: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 253: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 253: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 254: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 254: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0] Epoch: 255: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 255: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 256: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 256: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 257: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 257: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0] Epoch: 258: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 258: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 259: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 259: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 260: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 260: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 261: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 261: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 262: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 262: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 263: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 263: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 264: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 264: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 265: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 265: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 266: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 266: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0] Epoch: 267: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 267: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 268: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 268: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 269: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 269: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 270: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 270: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 271: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 271: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 272: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 272: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 273: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 273: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 274: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 274: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 275: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 275: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 276: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 276: 100% 213/213 [00:19<00:00, 11.01it/s, loss=0] Epoch: 277: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 277: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 278: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 278: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 279: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 279: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 280: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 280: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 281: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 281: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 282: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 282: 100% 213/213 [00:19<00:00, 11.01it/s, loss=0] Epoch: 283: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 283: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0] Epoch: 284: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 284: 100% 213/213 [00:19<00:00, 11.01it/s, loss=0] Epoch: 285: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 285: 100% 213/213 [00:19<00:00, 10.95it/s, loss=0] Epoch: 286: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 286: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0] Epoch: 287: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 287: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0] Epoch: 288: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 288: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0] Epoch: 289: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 289: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 290: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 290: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0] Epoch: 291: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 291: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 292: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 292: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 293: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 293: 100% 213/213 [00:19<00:00, 11.06it/s, loss=0] Epoch: 294: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 294: 100% 213/213 [00:19<00:00, 11.03it/s, loss=0] Epoch: 295: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 295: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0] Epoch: 296: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 296: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 297: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 297: 100% 213/213 [00:19<00:00, 11.05it/s, loss=0] Epoch: 298: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 298: 100% 213/213 [00:19<00:00, 11.02it/s, loss=0] Epoch: 299: 0% 0/213 [00:00<?, ?it/s]/content/ReCOGS/model/encoder_decoder_hf.py:828: FutureWarning: Version v4.12.0 introduces a better way to train encoder-decoder models by computing the loss inside the encoder-decoder framework rather than in the decoder itself. You may observe training discrepancies if fine-tuning a model trained with versions anterior to 4.12.0. The decoder_input_ids are now created based on the labels, no need to pass them yourself anymore. warnings.warn(DEPRECATION_WARNING, FutureWarning) Epoch: 299: 100% 213/213 [00:19<00:00, 11.04it/s, loss=0] Epoch: 100% 300/300 [1:37:02<00:00, 19.41s/it] INFO:root:Training is finished ... Iteration: 100% 24/24 [00:17<00:00, 1.36it/s, acc=1] Iteration: 100% 165/165 [32:14<00:00, 11.72s/it, acc=0.91] obj_pp_to_subj_pp: 12.1 cp_recursion: 61.1 pp_recursion: 64.3 subj_to_obj_proper: 90.0 prim_to_obj_proper: 85.6 prim_to_subj_proper: 99.9 LEX: 99.91333333333333 OVERALL: 91.03333333333333
moved this last one to new notebook to run in parallel, https://colab.research.google.com/drive/13FRQeAjyPOhBtTdrpW8caL25rNryLn5-?authuser=0#scrollTo=VxRXS4jinmeD
In [ ]:
!python run_cogs.py --model_name ende_transformer --gpu 1 --train_batch_size 128 --eval_batch_size 128 --lr 0.0001 --data_path ./recogs_positional_index --output_dir ./results_recogs_positional_index --lfs cogs --do_train --do_test --do_gen --max_seq_len 512 --output_json --epochs 300 --seeds "43;67;78;89;100" # paper's seeds, not sure how they chose them
papers seeds + 1, for runs 6-10 inclusive out of 10
results from other notebook ( https://colab.research.google.com/drive/13FRQeAjyPOhBtTdrpW8caL25rNryLn5-#scrollTo=VxRXS4jinmeD ) were:
seed 43: obj_pp_to_subj_pp 16.1%
seed 67: obj_pp_to_subj_pp 22.3%
seed 78: obj_pp_to_subj_pp: 23.7%
seed 89: obj_pp_to_subj_pp: 20.9%
seed 100: obj_pp_to_subj_pp: 14.0%
combine with earlier seeds in this notebook ( https://colab.research.google.com/drive/12mXX5L1I4rpwl1Jk8hCm-xyAkqiKJEo7 ):
seed 42: obj_pp_to_subj_pp: 23.8%
seed 66: obj_pp_to_subj_pp: 20.6%
seed 77: obj_pp_to_subj_pp: 23.2%
seed 88: obj_pp_to_subj_pp: 16.4%
seed 99: obj_pp_to_subj_pp: 12.1%
wu et al 2023 baseline with 4 layers (instead of 2) (not expected to be better):
>>> import numpy as np
>>> # not testing my own model, this is the wu et al 2023 baseline
wu_et_al_baseline_2023_4_layers_obj_pp_to_subj_pp = np.array([16.1,22.3,23.7,20.9,14.0,23.8,20.6,23.2,16.4,12.1])
>>> >>> wu_et_al_baseline_2023_4_layers_obj_pp_to_subj_pp.mean()
19.31
>>> wu_et_al_baseline_2023_4_layers_obj_pp_to_subj_pp.std()
4.082266527310533
>>> len(wu_et_al_baseline_2023_4_layers_obj_pp_to_subj_pp)
10
>>> import math
>>> wu_et_al_baseline_2023_4_layers_obj_pp_to_subj_pp_stderr = wu_et_al_baseline_2023_4_layers_obj_pp_to_subj_pp.std()/math.sqrt(10)
>>> (wu_et_al_baseline_2023_4_layers_obj_pp_to_subj_pp.mean() - wu_et_al_baseline_2023_4_layers_obj_pp_to_subj_pp_stderr*1.96, wu_et_al_baseline_2023_4_layers_obj_pp_to_subj_pp.mean() + wu_et_al_baseline_2023_4_layers_obj_pp_to_subj_pp_stderr*1.96)
(16.77978499253522, 21.840215007464778)
current draft reports (on n=5 instead of n=10):
19.22 +/- 4.4128902093752576 (sample mean +/- std)
95% confidence interval 15.4% to 23.1%
we update at n=10 to:
19.31 +/- 4.082266527310533 (sample mean +/- std)
95% confidence interval 16.8% to 21.8%
In [ ]: